MASP-1 and MASP-2 Do Not Activate Pro–Factor D in Resting Human Blood, whereas MASP-3 Is a Potential Activator: Kinetic Analysis Involving Specific MASP-1 and MASP-2 Inhibitors

The complement system can be activated by three interconnected routes: the classical, lectin, and alternative pathways (1). Activation of both the classical and lectin pathways leads to the formation of the same C3 convertase, namely C4bC2a. Upon C3 cleavage, C3b is generated that forms a transient complex with factor B (FB). The FB component of the proconvertase C3bB is then cleaved by factor D (FD) to produce the alternative pathway C3 convertase, C3bBb. Hence, the alternative pathway serves as an important amplification loop for both the classical and lectin pathways; however, it can be also activated on its own by the tick-over mechanism (1).

Components of the lectin pathway are (at least) five different pattern recognition molecules: mannose-binding lectin (MBL), H-ficolin (ficolin-3), l-ficolin (ficolin-2), M-ficolin (ficolin-1), and CL-LK, three MBL-associated serine proteases (MASP-1, MASP-2, and MASP-3), and two MBL-associated proteins, MAp19 (sMAP) and MAp44 (MAP-1) (2–5). MASP-1 was shown to be the exclusive initiating enzyme during the activation of the lectin pathway (6, 7), and, together with MASP-2, they are both essential for the C4bC2a convertase formation. In contrast, the function of MASP-3 has been unclear. MASP-1 has relatively broad substrate specificity (8), and it was implicated that it has a role in several proinflammatory reactions (9, 10) and coagulation (11). Both MASP-1 and MASP-3 were suggested to participate in pro-FD conversion to FD (12, 13), and it was speculated that even the proenzyme form of MASP-3 might be able to perform this process (13).

FD is a key enzyme for the alternative pathway. Its sole natural substrate is FB in complex with C3b (14). When purified from blood, only the active form can be retrieved (15, 16), and it is accepted that FD is present predominantly in the active form in normal blood (15), although some pro-FD was reported to be present in diisopropyl-fluorophosphate–treated human plasma (17). It was speculated that FD might be activated before secretion by an unknown protease within the cells (18). In most cell types, expression of natural (19) or recombinant pre–pro-FD (i.e., a construct containing a signal peptide and the putative propeptide) resulted in accumulation of the active form in the cell-culture supernatant (20), although a portion of the produced FD was proenzymic (20). In contrast, pro-FD was successfully produced in insect cells (21). In vitro, trypsin was found to be the best activator of pro-FD, whereas thrombin, kallikrein, and plasmin can also activate it with varying efficiencies (21); however, the exact kinetic constants have not been determined. In MASP-1/3(^−/−) mice, only pro-FD was found (12), and at first, it was suggested that MASP-1 might be the enzyme responsible for pro-FD cleavage. However, reconstitution of the MASP-1/3(^−/−) mouse serum with recombinant MASP-1 did not restore the alternative pathway (12). Later, MASP-3 was also implicated as a possible pro-FD activator (13), suggesting a new role for this enzyme. In contrast, alternative pathway activity was observed in MASP-1/3(^−/−) human serum, suggesting that neither MASP-1 nor MASP-3 is essential in pro-FD conversion to FD (22). Nevertheless, this scientific debate brought to light some questions that for a long time have seemed to be resolved (23, 24). Interestingly, in Cfh(^−/−)MASP-1/3(^−/−) mice (deficient for factor H, MASP-1, and MASP-3), uncontrolled alternative pathway activation was present just like in Cfh(^−/−) mice, suggesting that MASP-1 and -3 are not required for alternative pathway activity in Cfh(^−/−) mice (25). In summary, the exact roles of MASPs in pro-FD activation have not been clarified yet, and additionally, we cannot exclude the possibility that significant differences may exist between human and mouse complement regarding pro-FD activation.

In this study, we aimed to clarify the role of MASPs in pro-FD activation in human blood. For this purpose, we determined pro-FD cleavage rates using: 1) purified enzymes; 2) normal human serum and plasma in the absence and presence of specific MASP inhibitors; and 3) normal human plasma with exogenous MASPs added. Our quantitative data (in combination) rule out all MASP proenzymes, as well as active MASP-1 and MASP-2 as direct pro-FD activators in resting blood; however, active MASP-3 is a likely physiological activator. The term “resting blood” in this paper refers to blood, in which neither the coagulation system nor the complement system is activated above the baseline level.

Wild-type rMASP-1, MASP-2, and MASP-3 catalytic fragments (cf) encompassing the complement control protein domain 1- and -2-serine protease domains (SP) (MASP-1cf, MASP-2cf, and MASP-3cf) were produced as described (7, 26, 27). MASP-1cf and MASP-2cf are obtained as active enzymes, whereas MASP-3cf is produced as a zymogen. Zymogen MASP-3cf was further purified on a YMC-Q30 (YMC Europe) anion-exchange column (16 × 100 mm) by applying a 20-column–volume gradient of 60–200 mM NaCl in 10 mM Tris-HCl (pH 8.2). Active MASP-3cf was produced from the zymogen by cleavage using MASP-1cf, and then MASP-1cf was chromatographically removed as described (28). Active MASP-3cf was further purified on a YMC-Q30 column (YMC Europe) the same way as described above for the proenzyme. High-purity active MASP-3cf was eluted as the first major peak at ∼150 mM NaCl. The stable zymogen variants MASP-1cf R448Q, and MASP-2cf R444Q, were prepared as described (7, 29). Human anti-thrombin (activity ≥2800 NIH units/mg protein; catalog number T-1063) and bovine cationic trypsin (activity ∼10,000 BAEE units/mg protein; catalog number T-8003) were purchased from Sigma-Aldrich. FB and FD from human plasma were purchased from Merck (Calbiochem), and C3b was a gift from Prof. Mohamed R. Daha (Leiden University, Leiden, The Netherlands). The MASP-1–specific SGMI-1 and the MASP-2–specific SGMI-2 inhibitors were prepared according to Héja et al. (30). Blood was drawn from seven healthy volunteers into Vacutainer serum tubes (BD Diagnostics) to produce fresh serum and into Vacutainer citrate tubes (BD Diagnostics) to produce fresh plasma. After centrifugation, serum samples were pooled and kept frozen in aliquots at −70°C to produce normal human serum. Human plasma was also prepared by centrifugation, pooled, and kept at −70°C in aliquots. Cy3-NHS ester and Cy5-NHS ester were purchased from GE Healthcare. Acetonitrile and ultrapure (liquid chromatography mass spectrometry [MS]–grade) water for mass spectrometric analysis were purchased from Merck (Darmstadt, Germany).

The C-terminally poly-histidine–tagged (His₆) human pre–pro-FD encoding DNA was purchased as a synthetic gene from Entelechon (Regensburg, Germany). The coding sequence was manually codon optimized for expression in Spodoptera frugiperda 9 cell line (Sf9) cells taking into account the S. frugiperda codon usage frequencies found in the Codon Usage Database (http://www.kazusa.or.jp/codon/). The clone encodes the human signal peptide, activation peptide, and mature human FD polypeptide chain followed by a His₆ tag. The gene sequence and the translation are provided in Supplemental Table I. The PstI-EcoRI fragment of the synthetic gene was subcloned into the pVL1392 (BD Biosciences) transfer vector. The resulting pVL1392_hu-prepro-FD-H₆ plasmid was used to cotransfect Sf9 cells with linearized Autographa californica nuclear polyhedrosis virus baculovirus DNA using the BaculoGold transfection kit (BD Biosciences). The recombinant baculovirus was amplified by several rounds, and secreted protein was verified by Western blotting with a monoclonal anti-human complement FD Ab (catalog number GAU 010-04-02; Thermo Scientific). For large-scale expression of recombinant pro–FD-H₆ in Sf9 cells (BD Biosciences), a 100-ml cell suspension (cell density of 2 × 10⁶ cells/ml) in Insect-Xpress protein-free medium (Lonza) was infected with the recombinant baculovirus strain by adding 1.5 ml (∼2 × 10⁸ PFU/ml) virus stock. After 3.5 d of incubation at 27°C, the culture supernatant was collected by centrifugation (500 × g, 5 min) for purification.

The recombinant human pro-FD–H₆-containing supernatant was gel-filtrated on a Sephadex G-25 (GE Healthcare) 50 × 195-mm column in 10 mM imidazole and 250 mM NaCl (pH 7.5) buffer. Fractions containing the desired protein were combined, applied to a 16 × 60-mm Ni-NTA Superflow (Qiagen) column, and eluted with a linear gradient of 10–250 mM imidazole in a 250 mM NaCl and 50 mM NaH₂PO₄ (pH 7.5) buffer. The pro-FD–H₆-containing fractions were combined and dialyzed overnight excessively against 20 mM NaH₂PO₄ (pH 6.8) at 4°C, then pro-FD–H₆ was applied to a 16 × 125-mm YMC-BioPro S30 (YMC Europe) cation-exchange column equilibrated with 20 mM NaH₂PO₄ (pH 6.8) buffer, and eluted with a 50–350 mM NaCl linear gradient in the same buffer. During this chromatographic step, two separable forms of pro-FD–H₆ were observed. The first eluted peak was designated as “A,” and the second peak, the major form, was designated as “B.” Fractions were assayed for purity by SDS-PAGE, and the appropriate fractions of the two forms were combined separately, then concentrated on 10-kDa cutoff concentrators to ∼1 mg/ml, and stored frozen in aliquots. The “B” form represents intact human pro-FD–H₆ (see the next section), and the term “pro-FD” will refer to this form unless otherwise stated. The concentration of pro-FD was calculated based on the extinction coefficient ε₂₈₀ = 28 460 M⁻¹ cm⁻¹, and a molecular mass of 25.8 kDa.

Initial MALDI-TOF MS analysis exhibited that the “B” form is consistent with the m.w. of pro-FD–H₆, and the “A” form is degraded probably at the N terminus (data not shown). The “B” form of recombinant human pro-FD was further analyzed by liquid chromatography (HPLC) coupled to electrospray ionization (ESI) quadrupole time-of-flight (QTOF) MS. A total of 10 μl sample was desalted and separated without pretreatment using reversed phase chromatography on a Dionex U3000 RSLC system (Thermo Scientific, Dreieich, Germany). The separation was performed on a Poroshell 300SB C18 column (1 × 75 mm; 5-μm particle diameter, 300 Å pore size; Agilent Technologies) at 75°C and 200 μl/min applying a 40-min linear gradient with varying slopes. In detail, the gradient steps were applied as follows (min/% Eluent B): 0/5, 3/5, 5/15, 20/32, 22/36, 29/50, 31/100, 34/100, 36/5, and 40/5. Eluent B was acetonitrile with 0.1% formic acid, and solvent A was water with 0.1% formic acid. To avoid contamination of the mass spectrometer with buffer salts, the HPLC eluate was directed into waste for the first 2 min. Continuous MS analysis was performed using a QTOF mass spectrometer (Maxis UHR-TOF; Bruker, Bremen, Germany) with an ESI source operating in positive ion mode. Spectra were taken in the mass range of 600–2000 m/z. External calibration was applied by infusion of tune mix via a syringe pump during a short time segment at the beginning of the run. Raw MS data were lock-mass corrected (at m/z 1221.9906) and further processed using Data Analysis 4.1 and MaxEnt Deconvolution software tools (Bruker).

For producing active FD from its zymogen form, pro-FD was incubated with a 1/1000 amount of trypsin (by weight) at 37°C for 25 min in 20 mM NaH₂PO₄ and ∼300 mM NaCl (pH 6.8) buffer. The reaction was attenuated by 10-fold dilution with ice-cold equilibration buffer, then the sample was applied immediately to a 5 × 58-mm SP Sepharose High Performance (GE Healthcare) column equilibrated with 20 mM NaH₂PO₄ and 200 mM NaCl (pH 6.8) buffer, and eluted with a 200–550 mM NaCl linear gradient in the same buffer. Fractions were assayed for purity by SDS-PAGE, then selected fractions were combined and concentrated on 10 kDa cutoff concentrators to ∼0.5–1 mg/ml, and stored frozen in aliquots. The concentration of FD was calculated based on the extinction coefficient ε₂₈₀ = 28,460 M⁻¹ cm⁻¹ and a molecular mass of 25.2 kDa.

A total of 100 μg/ml C3b and 100 μg/ml FB were incubated with 50 μg/ml recombinant FD or plasma FD in 100 mM NaCl, 50 mM Tris-HCl, and 10 mM MgCl₂ (pH 7.5) at 37°C. Samples were taken at 0, 0.5, 1, 2, 4, 6, and 24 h. Reactions were stopped by 2-fold dilution with SDS-PAGE sample buffer and heating for 2 min at 95°C. Samples were then analyzed by SDS-PAGE under reducing conditions.

Pro-FD was dialyzed overnight excessively against 140 mM NaCl, 50 mM HEPES, and 0.1 mM EDTA (pH 7.4), then concentrated to 0.5–1 mg/ml and stored frozen in aliquots. The cleavage assays were performed in the same buffer at 10 or 20 μM final pro-FD concentration, and the samples were incubated for up to 3 h at 37°C. Trypsin, thrombin, active MASP-1cf, active MASP-2cf, active MASP-3cf, zymogen MASP-1cf R448Q, zymogen MASP-2cf R444Q, and zymogen MASP-3cf were applied at 0.20 nM, 635–955 nM, 33–68 nM, 13.5 nM, 30–61 nM, 7.0 μM, 6.5 μM, and 5.3–9.4 μM final concentrations, respectively, and samples were taken every 30–40 min. The reaction was stopped by 30-fold dilution with ice-cold equilibration buffer, then the sample was applied immediately to a 4.6 × 50-mm YMC-BioPro SP-F nonporous high-pressure cation exchange column (YMC Europe GmbH) equilibrated with 20 mM NaH₂PO₄, 100 mM NaCl (pH 6.8) buffer, and eluted with a 100–450 mM NaCl linear gradient in the same buffer at 0.8 ml/min ensuring rapid analysis. The observed first-order rate constant (k_obs) was determined by nonlinear regression using the area under the curve (AUC) = AUC_o × exp (− k_obs × t) equation, in which AUC is the integrated area of the pro-FD peak, and AUC_o is the area at the zero time point. In order to make the rate constants of the different reactions comparable, k_obs/[E]_T values were calculated, in which [E]_T is the total enzyme concentration. The k_obs/[E]_T value can be considered as an approximation of the catalytic efficiency (k_cat/K_M) according to the Michaelis-Menten kinetics when the substrate concentration is lower than the K_M.

For fluorescent labeling of pro-FD and FD, we used Cy3 and Cy5 CyDye monoreactive NHS Esters (GE Healthcare). Dried dyes were dissolved to 1 mg/ml in anhydrous DMF (Merck) and stored frozen in aliquots. Labeling reactions were performed in 20 mM NaH₂PO₄ and ∼300 mM NaCl (pH 6.8) buffer with 10:1 dye to protein molar ratio and incubated in the dark for 4 h at room temperature and then overnight at 4°C. The reactions were stopped by adding 1/10 volume 1 M Tris-HCl (pH 8) buffer, and after 30-fold dilution with ice-cold equilibration buffer, samples were applied to a 5 × 55-mm YMC-BioPro S30 (YMC Europe) column equilibrated with 20 mM NaH₂PO₄ (pH 6.8) buffer and eluted with a 0–350 mM NaCl linear gradient in the same buffer. Single-labeled Cy3–pro-FD had an A₅₄₉/A₂₈₀ ratio of ∼4, and single-labeled Cy5-FD had an A₆₄₉/A₂₈₀ ratio of ∼6. Fractions were assayed homogeneity by SDS-PAGE, and then selected fractions containing single-labeled proteins were combined and concentrated on 10 kDa cutoff concentrators to 1 mg/ml and stored frozen in aliquots. The concentrations of Cy3–pro-FD and Cy5-FD were calculated based on the extinction coefficients of Cy3 (ε_549nm = 150,000 M⁻¹cm⁻¹) and Cy5 (ε_649nm = 250,000 M⁻¹cm⁻¹) and molecular masses of 26.5 or 25.9 kDa, respectively.

Conversion of Cy3-labeled pro-FD into Cy3-FD in serum or plasma was monitored as follows. A total of 5 μl 1-mg/ml Cy3–pro-FD was added to 50 μl normal human serum or citrated plasma, and the mixtures were incubated at 37°C for 24–26 h. Samples were withdrawn at every hour in the first 8 h, and then an overnight (24–26 h) sample was collected.

In order to test the contribution of the thrombin content of serum to convert pro-FD into FD, we added 1 ml serum to a Vacuette Hirudin Blood Tube (Roche) containing >45 μg hirudin. Cy3–pro-FD conversion was tested on the same batch of hirudin-treated and untreated serum as above.

In order to test the impact of MASP inhibitors on pro-FD conversion into FD, we added them to serum or plasma prior to the addition of labeled pro-FD. A total of 1 μl SGMI-1 (15 μM final concentration) was added to 50 μl normal human serum or citrated plasma, and the mixtures were incubated for 10 mins at room temperature, then 5 μl 1-mg/ml Cy3-labeled pro-FD was added, and the mixtures were incubated at 37°C for 24–26 h. Samples were withdrawn at every hour in the first 8 h, and then an overnight (24–26 h) sample was collected. A total of 1 μl appropriately diluted SGMI-2 (15 or 1 or 0.1 μM final concentration) was added to 50 μl normal human serum or citrated plasma, and the mixtures were incubated for 10 min at room temperature, then 5 μl 1-mg/ml Cy3-labeled pro-FD was added, and the mixtures were incubated at 37°C for 24–26 h. Samples were withdrawn at every 1 to 2 h in the first 8–14 h, and finally, an overnight (24–26 h) sample was collected.

In order to test the impact of exogenously provided MASPs on pro-FD conversion into FD, we added 1 μl appropriately diluted MASP-1cf, MASP-2cf, or MASP-3cf (50–200 nM final concentration) to 50 μl citrated normal human plasma, and the mixtures were incubated for 10 min at room temperature, then 5 μl 1-mg/ml Cy3-labeled pro-FD was added, and the mixtures were incubated at 37°C for up to 24 h. Samples were withdrawn at every hour in the first 7 h, and finally, an overnight (24–26 h) sample was collected, except when 200 nM MASP-3cf was used, in which samples were withdrawn at every 15 min for up to 2 h.

All reactions were stopped by 10-fold dilution with SDS-PAGE sample buffer and heating for 2 min at 95°C. Cy5-labeled FD was diluted to 3.6 ng/μl with SDS-PAGE sample buffer then treated for 2 min at 95°C. A total of 2 μl cleavage assay samples containing 18 ng Cy3-labeled protein were mixed with 5 μl diluted Cy5-labeled FD (18 ng), the mixed samples were run on SDS-PAGE under reducing condition, and the gels were scanned with a Typhoon laser scanner (GE Healthcare). Band intensities were quantified by densitometry. Because the two bands were partially overlapping, only the time point at which the intensity of the Cy3–pro-FD band equaled that of the Cy3-FD band could be reliably determined on a particular gel. This time point is considered to be the half-life of Cy3–pro-FD. When the intensity of the Cy3–pro-FD band was higher at a given time point and lower in the subsequent time point than the intensity of the Cy3-FD band, we used interpolation.

Human pre–pro-FD was expressed in insect cells using a baculovirus vector as described in the 2Materials and Methods. The construct contained the human signal sequence and a C-terminal His₆ tag to aid purification. As previously described, insect cells produce predominantly pro-FD (21). Under our expression conditions, we obtained two major variants eluting in different peaks termed “A” and “B” during the last cation exchange step. Preliminary MALDI-TOF analysis exhibited that the “B” form is consistent with the m.w. pro-FD-H₆, and the “A” form is probably degraded (data not shown). Further analysis by HPLC–ESI-QTOF MS (Fig. 1) showed that the m.w. of the “B” form is consistent with an N-terminal propeptide sequence of APPRGR. In all subsequent experiments, we used the “B” form of pro–FD-H₆, hereafter simply termed as pro-FD. Because our major goal was to detect pro-FD activation by various serine proteases, which involves cleavage at the N terminus, the small tag at the C terminus most likely does not interfere with this reaction. We also converted our pro-FD into active FD by trace amounts of trypsin and purified it to remove the trypsin activity. The produced active enzyme cleaved FB-C3b in a similar fashion as the purchased control FD purified from human plasma (data not shown).

Pro-FD can be very effectively activated by trypsin in vitro (21); therefore, we also used bovine cationic trypsin as a highly active control protease in our assay system. Samples of pro-FD (10–20 μM) were treated with trypsin, thrombin, MASP-1cf, MASP-2cf, MASP-3cf, and also the proenzymic forms of all three MASPs at 37°C. The cleaving enzymes were used at concentrations suitable to produce ∼80% consumption of pro-FD in <3 h, as determined from preliminary experiments. Aliquots containing 10 μg total pro-FD + FD were withdrawn at 30–40-min periods, diluted 30-fold with ice-cold binding buffer, and analyzed immediately on a small cation exchange column as described in the 2Materials and Methods. The peak corresponding to pro-FD gradually disappears, whereas a new peak with increasing intensity corresponding to FD appears over time (Fig. 2). Fitting the AUC values of the pro-FD peak versus time (Fig. 3) yielded k_obs values for each reaction. Normalized by the enzyme concentrations the cleavage efficiencies can be compared, as summarized in Table I. The highest cleavage rate, by far, is produced by trypsin, whereas active MASP-1cf, MASP-2cf, MASP-3cf, and thrombin are fairly efficient activators of pro-FD. Proenzymic MASPs have either marginal (R448Q MASP-1cf) or no activity at all.

In order to model pro-FD cleavage in human blood, we developed a novel experimental setup that is suitable to monitor pro-FD cleavage in a complex mixture. We labeled pro-FD with Cy3 dye and followed its cleavage in normal human serum or citrated plasma. In both cases, Cy3-labeled pro-FD was added at 3.5 μM final concentration, and its cleavage was followed at 37°C. Adding the concentrated labeled protein to serum or plasma caused only a minimal (10%) dilution of the original serum or plasma samples. Aliquots were withdrawn typically at every hour, then immediately diluted 10-fold with SDS-PAGE sample buffer, and heated for 2 min at 95°C to stop the reactions. Portions of the treated samples containing 0.18 μl original serum or plasma and ∼18 ng total Cy3–pro-FD + Cy3-FD were loaded per lane and analyzed by SDS-PAGE (Fig. 4). Consumption of the Cy3–pro-FD band was apparent over time, and a new band appeared with a slightly higher mobility. In most gels, the SDS-PAGE samples containing Cy3–pro-FD plus Cy3-FD (18 ng total) were mixed with samples containing 18 ng Cy5-FD to allow multiplexing and unambiguous distinction of the uncleaved and cleaved forms (Fig. 4A). Our results showed that Cy3–pro-FD was converted to Cy3-FD in serum with a t_1/2 of ∼3 h and in plasma with a t_1/2 of ∼5 h. The difference is probably due to thrombin and possibly other coagulation enzymes present in serum. Preliminary experiments with hirudin-treated serum revealed that the t_1/2 of Cy3–pro-FD increased by ∼30 min compared with normal serum (data not shown), hence the difference between serum and plasma can only be partially attributed to the impact of thrombin.

Most importantly, Cy3–pro-FD was converted into Cy3-FD in human plasma as well, suggesting that there is a proteolytic activity capable of activating pro-FD in human blood, even in the absence of active coagulation enzymes. The activator in plasma is likely to be the physiological pro-FD activator in resting blood in which activation of the coagulation cascade does not occur.

In order to assess whether any of the MASPs could be a potential physiologic pro-FD activator, based on the k_obs/[E]_T values obtained from in vitro cleavage experiments, we calculated corresponding MASP concentrations required to produce the observed pro-FD t_1/2s of ∼3 and 5 h (Table II). These data suggest that each MASP (31) could be considered as a potential activator, but only if a significant proportion is already activated.

Previously we have developed specific inhibitors for MASP-1 and MASP-2 (30). SGMI-1 is a specific MASP-1 inhibitor with an apparent inhibition constant (K_i) of ∼7 nM, whereas SGMI-2 is a highly selective MASP-2 inhibitor with a K_i of ∼6 nM. We observed that SGMI-2 also inhibits MASP-3, although very poorly, with a K_i of ∼5 μM (6). We used these inhibitors at various concentrations to assess their effects on pro-FD activation in serum and plasma.

SGMI-1 did not affect the conversion of Cy3–pro-FD to Cy3-FD even at the highest concentration (15 μM) applied (Fig. 5, Table III) neither in plasma nor in serum. This experiment unambiguously rules out MASP-1 as a direct pro-FD activator. Because in vitro active MASP-1 is a potent pro-FD activator, these results also imply that there is no significant amount of active MASP-1 in resting blood.

At 100 nM, which is well above the K_i for MASP-2, SGMI-2 had no effect at all, and even 1 μM SGMI-2 had only small effect (Fig. 5, Table III). This ruled out MASP-2 as well as direct pro-FD activator. Although we applied SGMI-2 primarily because it inhibits MASP-2, calculations presented in Table II would imply that essentially all MASP-2 would have to be in the active form to produce the measured Cy3–pro-FD half-life in plasma. This is in contrast with the observation that MASP-2 circulates as a proenzyme in the blood (32). So the results are coherent with all calculations and previous observations.

However, to our surprise, at high (15 μM) inhibitor concentration, SGMI-2 significantly slowed down Cy3–pro-FD conversion both in serum and plasma (Fig. 5, Table III). The effect was more pronounced in plasma. The difference between serum and plasma can be explained by the effect of thrombin and possibly other coagulation enzymes, which are not inhibited by SGMI-2 (6). In contrast, the measured concentration dependence is highly consistent with the K_i value of 5 μM SGMI-2 for MASP-3 (6).

In summary, our experiments so far rule out MASP-1 and MASP-2 as physiological pro-FD activators in resting blood, suggesting that some other protease or proteases are responsible for this activity. Our observations are consistent with active MASP-3 to be such an enzyme, but also imply that for such a role, a significant portion of MASP-3 should be in active form even in resting blood. These assumptions require further experiments to prove unambiguously.

In order to further characterize the effect MASPs on pro-FD conversion into FD, we added MASP-1cf, MASP-2cf, or MASP-3cf to normal human plasma at 50–200 nM concentrations. Because we have shown before that only the active forms have considerable activity toward pro-FD (Tables I, II), we used active MASPs. At 200 nM, final plasma concentration exogenously provided MASP-1 and MASP-2 did not reduce the half-life of Cy3–pro-FD at all, despite the fact that both can cleave pro-FD in a purified system. This suggests that both active enzymes are rapidly inactivated probably by C1 inhibitor and antithrombin. In contrast, MASP-3 reduced the half-life of Cy3–pro-FD significantly in a dose-dependent fashion (Fig. 6, Table IV). At 50 nM MASP-3cf, the t_1/2 was reduced to 2 h, whereas at 200 nM to <30 min, suggesting that MASP-3 is not inhibited and indeed capable of cleaving pro-FD in the complex matrix of human plasma.

Summing up, the experiments in the presence of exogenously added MASPs reinforced our assumption that MASP-3 can indeed convert pro-FD into FD in plasma, and among lectin pathway proteases only MASP-3 can be considered as the pro-FD activator of resting blood (Fig. 7).

Controversy in the literature regarding the role of the lectin pathway, in particular the involvement of MASP-1 and MASP-3 (12, 13, 22), in the activation of the alternative pathway led us to take a closer look at the impact of MASPs on pro-FD cleavage. For this purpose, we already had highly purified recombinant catalytic fragments of all MASPs, both in proenzymic and active form, and highly selective inhibitors for MASP-1 and MASP-2.

We also needed pro-FD that we produced in the baculovirus insect cell expression system. It had been previously reported that in this expression system, pro-FD is released into the culture medium in two zymogen forms with respective activation peptides AAPPRGR and APPRGR (21) in a 2:1 ratio. The high mass resolution and accuracy attained by the ESI-QTOF approach enabled us to unambiguous identify the produced proteins. In our experiments, the vast majority of pro-FD purified from insect cell culture medium contained the APPRGR activation peptide. This finding is in agreement with the predicted signal peptide cleavage site (33). Notably, contrary to what was found earlier, the other form having an additional N-terminal alanine (AAPPRGR) could be barely detected. It should be noted, however, that Yamauchi et al. (21) seem to have used a shortened signal peptide, whereas we used the natural human signal peptide. So apparently APPRGR represents the dominant natural propeptide.

In order to compare their inherent capacity to activate pro-FD first, we have determined the cleavage rate constants for all MASPs. Interestingly, there was only a slight difference among the active MASPs indicating that, at least in vitro, they are all capable of activating pro-FD. The highly purified zymogen forms had zero (MASP-2 and MASP-3) or marginal (MASP-1) pro-FD cleaving activity. These results indicate that the main question is not which MASPs have the ability to cleave pro-FD, but rather which ones have sufficiently high concentration in the active form to carry out such cleavage in resting blood. In contrast, during lectin pathway activation, when all of the three MASPs are present in their active form, activation of pro-FD may be accelerated considerably.

In order to model the physiologic situation, we measured the half-lives of labeled pro-FD in normal human plasma and serum. The t_1/2 was shorter in serum, indicating that coagulation enzymes, including thrombin, may be able to convert pro-FD into FD. In complement assays, serum is used routinely. Our results indicate that one should be careful when drawing conclusion on the natural pro-FD activating capacity of an individual or model animal by using serum samples (23, 24), as low-level activation of pro-FD might occur by backup enzymes in the serum even in the absence of the natural pro-FD activator of the resting blood. In this respect, it is noteworthy that FD-deficient mouse serum reconstituted to <1% of the normal FD level restored 23% of the alternative pathway activity (34), suggesting that the normal level of FD is not limiting, as opposed to previous results (15).

In resting blood, the coagulation cascade is not activated, and therefore plasma experiments may represent a better model for the normal physiological pro-FD conversion than serum experiments. Because labeled pro-FD was converted to FD in plasma with a t_1/2 of ∼5 h, it is reasonable that a proteolytic activity exists in resting blood that converts pro-FD to FD. The labeled pro-FD concentration that we used was 3.5 μM, because such high concentration was necessary for reliable detection. This value is significantly higher than the physiological FD concentration (80 nM) (15), distorting unfavorably the enzyme to substrate ratio, therefore it is plausible that the pro-FD conversion rate in the circulation is higher, providing a significantly shorter t_1/2 than the measured ∼5 h. Another contributing factor could be the “buffer effect” of citrated plasma. It is also possible that in the presence of Ca²⁺, the main converting enzyme has a higher activity. In contrast, in plasma, further activation of lectin pathway proteases and coagulation enzymes above the baseline level can be excluded; therefore, we used both serum and plasma to evaluate the effects of our MASP inhibitors.

When we added an MASP-1–specific inhibitor (SGMI-1; K_i ∼7 nM) to serum or plasma at 15 μM final concentration, the pro-FD conversion rate was no different from the inhibitor-free values. It is noteworthy that SGMI-1 inhibits both active (30) and proenzymic MASP-1 (7). Based on these observations, we unambiguously ruled out MASP-1 as a direct pro-FD activator in resting blood. If MASP-1 is somehow involved in pro-FD maturation, its contribution can only be indirect (e.g., through MASP-3 activation) (7, 13).

To our surprise, SGMI-2, a highly selective MASP-2 inhibitor with a K_i of 6 nM, did slow down pro-FD conversion both in serum and plasma, although only at high inhibitor concentration (15 μM). The effect was more pronounced in plasma. In this study, we became suspicious, because according to our calculations, essentially all MASP-2 would have to be in the active form to produce a 5-h t_1/2 for labeled pro-FD in plasma. This is in sharp contrast with previous observations that MASP-2 exists in the proenzyme form in resting blood (32). Importantly, SGMI-2 also inhibits MASP-3 (6), but only quite poorly with a K_i of ∼5 μM. When we examined how pro-FD cleavage rate in plasma depends on SGMI-2 concentration, we found that 1 μM SGMI-2 had a minor effect, and 100 nM SGMI-2 had no effect at all. This concentration dependence was more consistent with the involvement of an enzyme inhibited by SGMI-2 with a K_i value in the micromolar range. We note that MASP-3 is such an enzyme. Nevertheless, based solely on these results, we cannot yet identify MASP-3 as the exclusive pro-FD activator in resting blood. Experiments with exogenously added MASPs further support the idea that MASP-3 is the physiological pro-FD activator, as we demonstrated that among MASPs only MASP-3 was capable of reducing the half-life of labeled pro-FD in plasma.

Our observations support the view of Takahashi et al. (24, 35) and Iwaki et al. (13), who proposed that MASP-3 acts as a pro-FD activator. In contrast to Iwaki et al. (13), who stated that even zymogen MASP-3 might activate pro-FD, we have shown that only active MASP-3 can carry out that cleavage. When Dahl et al. (36) first isolated MASP-3–containing complexes, the majority of MASP-3 appeared to be in the active (two-chain) form, even though they used protease inhibitors during the isolation. It is, however, not clear that activation of MASP-3 occurred during the isolation, or MASP-3 associated with MBL was already active. Our experiments suggest that for MASP-3 to be the major pro-FD activator, a significant portion (at least ∼10 nM) has to be in the active form in plasma to provide the observed cleavage rate. It will require further investigations to accurately determine what percentage of MASP-3 is actually present in resting blood in the active form. It is also an intriguing question how zymogen MASP-3 is activated in the absence of a lectin pathway–activating surface. One possibility is that fluid-phase activation of lectin pathway proteases occurs slowly at a constant rate, but active MASP-1 and MASP-2 are rapidly inactivated by their natural inhibitors like C1 inhibitor and antithrombin (37). This assumption is supported by the presence of MASP-1–C1-inhibitor and MASP-1–antithrombin complexes in normal serum and plasma (38). Importantly, as active MASP-3 is not inhibited by C1 inhibitor (39) or any other identified serpin, it could slowly accumulate in such a scenario. Our observations with exogenously added MASPs support such a mechanism. When recombinant active MASP catalytic fragments were added to plasma, only MASP-3 reduced the t_1/2 of labeled pro-FD, indicating that whereas MASP-1 and -2 are inactivated rapidly, MASP-3 remains active.

Taken together, we have shown that in vitro, the active serine proteases generated during activation of the coagulation (tested with thrombin) and complement cascades (MASP-1, -2, and -3) can cleave pro-FD with comparable efficiencies. In serum, the contribution of the coagulation proteases may be significant. Furthermore, in resting blood, MASP-1 and MASP-2 cannot be direct activators of pro-FD.

In contrast, we have unambiguously demonstrated that at least one protease capable of activating pro-FD exists in normal human plasma. Although further experiments are needed to unequivocally identify the physiological and perhaps exclusive pro-FD activator in resting blood, in this study, we provided important evidences that activated MASP-3 is the likely candidate (Fig. 7).

We thank Dr. Katalin Kékesi (Eötvös Loránd University, Budapest, Hungary) for help with the Typhoon laser scanner and Prof. Mohamed R. Daha (Leiden University, Leiden, The Netherlands), who provided C3b. We also thank Dorina Rákóczi (Pázmány Péter Catholic University, Budapest, Hungary) for help during MASP-3cf purification.

The authors have no financial conflicts of interest.