MBL-associated serine protease
Fast protein liquid chromatography
Mannose-binding lectin (MBL) activates complement through MBL-associated serine proteases (MASP). A deficiency in MBL due to mutations at exon 1 of the human MBL gene is reported to cause vulnerability to infection. We examined sera of known MBL genotype by gel filtration and assessed their elution patterns using an ELISA for MBL and identified two MBL forms, a high-molecular-mass form and a lower-molecular-mass form. By the identification of either or both forms in individual sera, three types of patterns emerged: type 1 consisted of a high-molecular form; type 2, of a low-molecular form; and type 3, of both forms. Types 1, 2 and 3 corresponded, respectively, to a wild type (A/A), a homozygous mutation at codon 54 (B/B) and their heterozygote (A/B). One exception was a heterozygous LXPA/LYPB phenotype that exhibited the type-2 pattern. Binding to mannan and MASP-1/3 occurred exclusively with the high-molecular form. An apparent MBL deficiency does not in fact representa deficiency in MBL molecules but rather the presence of circulating oligomeric mutant MBL with impaired function.
Mannose-binding lectin (MBL) is produced in hepatocytes and secreted into plasma. It plays an important role in activation of complement in cooperation with MBL-associated serine protease (MASP)-1, MASP-2 and MASP-3 and works to exclude microorganisms in concert with phagocytic cells 1–3. A deficiency in MBL can cause a vulnerability to infection 4–6. It is well known that there are three point mutations in codons 52, 54 and 57 in exon 1 of the human MBL structural gene encoding collagen-like sequences 7–9. These mutations are thought to interfere with the maintenance of a stable quarternary structure and to promote degradation in the circulation 10–12 or to impair secretion 11, 13, leading to the MBL deficiency. Among Japanese subjects, only one mutation at codon 54 has been reported 14–18. Polymorphisms in the promoter region or in the untranslated region of the MBL gene also affect the level of MBL 19. Nucleotide substitutions at position –550 (G to C), at position –221 (G to C) and at position +4 (C to T) give rise to H(G)/L(C), Y(G)/X(C) and P(C)/Q(T) variants, respectively. Promoter haplotypes HY, LY and LX lead to high, intermediate and low levels of MBL, respectively 19–25.
We have previously reported that MBL was detected in all Japanese sera examined and no MBL deficiency was found with our enzyme-linked immunosorbent assay (ELISA) system 26, 27. However, there have been a number of reports concerning MBL deficiencies in individuals of other countries 8, 9, 19, 20, 22, 28. At first, it was thought that Japanese have no MBL mutations 29. However, it was later shown that mutations exist among the Japanese as well 14–18. The present study was initiated to elucidate why MBL is detectable in individuals with an MBL mutation using our ELISA system. During the course of our studies, we found that our system could detect mutated MBL and a certain relationship emerged between the MBL genotype and molecular forms and serum levels of MBL.
2.1 Identification of two molecular forms of MBL in serum
We examined a number of sera of known MBL genotype by means of gel filtration on Superose using fast protein liquid chromatography (FPLC) and assessed MBL elution patterns using our ELISA system. Two MBL with different molecular masses were found in these sera, one of 750 kDa (peak 1) and the other of 400–450 kDa (peak 2). Either or both peaks were identified in individual sera, and sera were classified as one of three types of elution patterns (Fig. 1). Type 1 consisted mainly of a 750-kDa MBL (peak 1) and type 2 mainly of a 400–450-kDa MBL (peak 2). Type 3 consisted of both peak 1 and peak 2. Type 1 had high levels of MBL, type 2 had very low MBL levels, and type 3 levels were intermediate (Fig. 1).
When these types were analyzed for genotype, types 1, 2 and 3 corresponded, respectively, to a wild type (A/A), a homozygote with a mutation at codon 54 (B/B) and a heterozygote (A/B). Polymorphisms at H/L, X/Y and P/Q did not have an impact on the pattern of the three types, except on that of LXPA/LYPB, which is heterozygous at A/B and X/Y. This genotype exhibited a very small peak 1 and a normal peak 2, leading to a type-2 classification.
As can be seen in Fig. 1, although types 1 and 2 are characterized by a single peak, either peak 1 or peak 2, type 1 also has a subtle peak 2 besides a large peak 1, and type 2 has a subtle peak 1 in addition to a small but distinct peak 2. Thus, all three types have two peaks with different ratios to each other. The ratio of peak 1 (the total amount of MBL in fractions 22 and 23) to peak 2 (MBL in fractions 28 and 29) was calculated for each genotype. It was found that the peak 1/peak 2 ratios in individual genotypes belonging to type 1 were roughly the same, and the mean ratio was calculated to be 14.4 (Table 1). As for type 2, the mutant homozygote LYB/LYB and the heterozygote LXA/LYB exhibited the same ratio of 0.08. The ratio in type 3 was distinguished by two apparently different values, 0.93 and 0.44. It is noteworthy that among the A/B heterozygotes, including HYPA/ LYPB, LYPA/LYPB and LXPA/LYPB, the ratios (0.93, 0.44, 0.08) declined in the order of HY, LY and LX, which represent the highest-, intermediate-, and the lowest-producing haplotype, respectively 19–25.
2.2 Relationship between the serum MBL level and MBL genotypes
Serum levels of MBL, as determined by our ELISA system for individuals whose genotype was known, were plotted against each respective genotype. A correlation was observed between the mean serum MBL level and genotype (Fig. 2). As can be seen in Fig. 2, the serum MBL level was highly dependent on mutation of the structural genes, A and B, moderately dependent on a polymorphism at Y and X, and slightly on one at H and L in the promoter region. A PQ polymorphism did not have a significant impact on the MBL level.
2.3 Binding to mannan and MBL-MASP-1/3 complex formation with the two forms of MBL
The mannan-binding activity of MBL was assessed after elution from the FPLC column. As shown in Fig. 3B, almost all activity was found at a position corresponding to peak 1. The identification of the MBL-MASP-1/3 complex was also determined after the elution (Fig. 3C). The MBL-MASP-1/3 complex was eluted exclusively in peak 1. In the presence of EDTA and 1 M NaCl, the complex was not detectable, indicating that complex formation was dependent on Ca2+ and ionic strength, as reported previously 30.
2.4 Changes in the MBL elution pattern on Superose column with and without Ca2+
Individual sera representative of the three types of MBL patterns were passed through an FPLC Superose column using buffers with and without Ca2+, and changes in elution patterns were compared. As shown in Fig. 4, the addition of Ca2+ to the elution buffer resulted in a complete disappearance of peak-1 MBL in type-1 serum, whereas in type-2 serum almost all peak-2 MBL eluted, although at a slightly retarded position. In type 3, while the peak-2 MBL eluted without a decrease, the peak-1 MBL decreased to a half, although both peaks eluted at slightly retarded positions. The latter finding indicated that MBL in peak 1 (high-molecular form) of type 3 was heterogeneous with respect to its agarose (SuperoseTM)-binding capacity.
2.5 Confirmation of MBL molecule in each peak
The preceding results suggest the question of whether peak 2 really represents MBL or another substance detectable by our MBL assay system. To clarify this point, FPLC fractions of peak 1 (fractions 22–24) and peak 2 (fractions 28–30) from the individuals of each type were each collected and rechromatographed on a monoclonal anti-MBL (3E7) antibody-coupled Sepharose 4B affinity column, and the proteins bound to this column were eluted with 0.1 M Gly-HCl (pH 1.8). These eluates were immediately neutralized, concentrated, reduced with 30 mM dithiothreitol in 8 M urea and analyzed by 12.5% SDS-PAGE and by Western blotting. MBL band detection was performed with a biotinylated mouse monoclonal anti-MBL antibody (3E7 or 131–1).
As shown in Fig. 5, both peak 1 and peak 2 exhibited MBL signals at 32 kDa with the monoclonal anti-MBL antibody 131–1. Type 1 had MBL in peak 1 only, type 2 in peak 2 only and type 3 in both peaks with almost equal intensity, confirming our previous results with gel chromatography. The same results were obtained with another monoclonal anti-MBL antibody, 3E7 (not shown). For further confirmation, the 32-kDa protein band on the polyvinylidene difluoride membrane of the peak-2 fraction was analyzed for its amino acid sequence, and the N-terminal 13-amino-acid sequence obtained was entirely identical with that reported for MBL.
In previous reports concerning the mutation at codon 54 and the level of MBL as determined by ELISA 8, 9, the MBL level in the wild type (A/A) was estimated to be around 1,600 μg/l, that in the heterozygote (A/B) at around 300 μg/l, and that in the homozygote of the mutation (B/B) at less than 10 μg/l. With our ELISA system, the mean level of MBL was 1,686 μg/l for the wild type, 636 μg/l for the heterozygote and 451 μg/l for the homozygote with a mutation at codon 54 (Table 1). It is noteworthy that, while their reported MBL levels for the wild type were almost identical to ours, those of the heterozygote and mutant homozygote were much lower. In our present gel filtration experiments, we identified two molecular forms of MBL, a high-molecular-mass form (peak 1) with capacities for mannan binding and MASP-1/3 complex formation, and a lower-molecular-mass form (peak 2) without those capacities. The wild type (A/A) consisted almost totally of a single large peak 1 with a trace peak 2, the heterozygote (A/B) was represented by two small peaks 1 and 2 of almost equal size, and the mutant homozygote (B/B) was composed of almost a single small peak 2 with a subtle peak 1.
From these results, one can presume that the previously reported MBL levels reflected only peak 1, while those in the present study included both peaks 1 and 2. To strengthen this assumption, we calculated the MBL levels in peak 1, using peak 1/peak 2 ratios and the mean MBL values of each genotype (Table 1). It was deduced that the MBL levels in peak 1 of the wild type, heterozygote and mutant homozygote were 1577, 254 and 33 μg/l, respectively, indicating that the values correspond well to the previously reported values. This in turn indicates that our ELISA detects MBL, regardless of a mutation at codon 54, and that two allelic forms of MBL are produced and present in the circulation. This is the reason why our ELISA system does not detect any so-called MBL deficiency in Japanese individuals.
The failure to detect mutated MBL or the lower-molecular-mass form of MBL (peak 2) in the previous reports could be due to differences in the antibodies employed in the ELISA 31. In our system, a polyclonal rabbit anti-MBL antibody was employed as antigen-capturing antibody and a monoclonal mouse anti-MBL antibody as detector. However, an identical monoclonal anti-MBL IgG was used for both antigen capture and antibody detection in ELISA systems used in most other studies 8, 9, 19–23, 25, 28, 31. A symmetrical ELISA using an identical monoclonal antibody against a repeating epitope in the MBL molecule is thought to be inadequate for detection of the oligomeric or allelic form of MBL 31. However, our ELISA system can measure total MBL including low-molecular-mass MBL and/or mutant MBL which would be ineffective in complement activation. In other words, ELISA procedures other than ours might measure only the active form of MBL. An ELISA with mannan as the MBL-capturing agent and anti-MBL antibody as the detector would also be inadequate for the detection of mutated MBL or lower-molecular-mass MBL, since they lack a mannan-binding capacity as substantiated in the present study and suggested in previous reports 31–33.
We applied pooled human serum with Ca2+-containing buffer to two sequentially connected affinity columns, first to a mannose-coupled Sepharose 6B affinity column and then to an anti-MBL (3E7)-coupled Sepharose 4B affinity column. Proteins trapped in the first column (mannose-coupled) and in the second column were then separately eluted with appropriate buffers and were each rechromatographed on a gel filtration column. It was revealed that proteins trapped in the first column (mannose-coupled) were exclusively peak-1 MBL and those in the latter column (anti-MBL) were peak-2 MBL (not shown). This indicates that a high-molecular-mass form of MBL (peak 1) binds to mannose-coupled Sepharose, while a low-molecular-mass form of MBL (peak 2) is likely to pass through such a column even in the presence of Ca2+. This phenomenon is consistent with findings of previous reports 31–33. So far, only a single peak at a high-molecular-mass position was reported for purified serum MBL in humans 34–37 and other species 37–39. It is simply thought that low-molecular-mass MBL (corresponding to peak 2) was overlooked because it did not bind to the carbohydrate resin, and only the fractions which did bind to the resin (corresponding to peak 1) were able to proceed to the next step in purification or identification.
Concerning the identification of the lower-molecular-mass MBL or mutated MBL in serum from individuals with different MBL genotype, only one report by Lipscombe et al. is available 31. They reported that MBL of wild type (A/A) was eluted from a Superdex 200 gel filtration column at a wide region peaking around 500 kDa (max. 900 kDa), while mutant type (B/B) was eluted at around 240–380 kDa. They analyzed those eluates by non-reducing SDS-PAGE and following Western blot with anti-MBL antibodies, and found that the wild-type MBL fraction consisted of a mixture of oligomers formed from two to eight subunits (each based on three identical 32-kDa chains) of apparent molecular masses around 200–700 kDa, with dimers and trimers constituting the predominant forms, while the mutant MBL fraction was predominantly composed of materials with 120–130 kDa, possibly corresponding to a monomer and/or an anomalously arranged monomer-like form comprising four 32-kDa chain subunits. As far as Lipscombe;apos;s data were taken into consideration, it would be conceivable that our peak 1 consists of a mixture of trimeric and dimeric MBL, and the peak 2 of monomer and/or monomer-like MBL.
Very recently, it was reported that there are two forms of MBL, MBL-I complex and MBL-II complex, each of which has a distinct MASP composition (MASP-1, 2 and 3) and biological activity 3. However, that study was directed only at the wild type of MBL (A/A), and the two forms were both derived from the MBL complexes isolated by affinity chromatography on mannan-Sepharose; and thus, the MBL-I and MBL-II complexes, also reported each as having different-molecular-mass MBL, do not correspond to our peak 2 and peak 1.
The disappearance of peak 1 in type 1 (wild type) on gel filtration with Superose under Ca2+-containing buffer indicates tight binding of the missing MBL to the agarose resin. MBL is known to bind to unmodified SepharoseTM40, 41, consisting of agarose. This binding was inhibited by 0.2 M mannose in the same Ca2+-containing buffer (not shown), indicating that the binding occurs through the carbohydrate recognition domain of MBL as previously reported 42. It was unexpected and interesting that around ahalf of peak 1 in type 3 (heterozygote) was eluted from the column. This indicates that MBL in peak 1 of the heterozygote, which should be the mixture of oligomers consisting of a wild-type subunitand mutant subunit at different ratios, exhibit different affinities to Superose in relation to their subunit constituents. Peak 2 of all types showed almost no binding to Superose, consistent withthe lack of mannan binding in peak 2 MBL.
It is well confirmed that higher oligomeric structures are necessary for MBL in its complement activation 2, 11, 32, 33, 37, 43–45. In the present study, the binding to MASP-1/3 and mannan was seen only with the larger-molecular form of MBL but not with the smaller one, which was consistent with this description. MASP-1 and MASP-2 were found to co-elute largely with the high-molecular-mass MBL (750 kDa) on Superose 6 gel filtration 46. Further,it was reported that MBL/MASP-2 complex, which was reconstituted between MBL and recombinant MASP-2, was eluted on gel filtration chromatography at the position corresponding to the larger-molecular form of MBL 47. Consequently, we extrapolate that MASP-2 should also be eluted in the peak 1 together with MBL/MASP-1/3 complex on gel filtration chromatography.
Our results show that the so-called MBL deficiency due to MBL mutation does not infer a deficiency of MBL molecules. Mutant MBL molecules are present in the circulation even in individuals with an apparent MBL deficiency, although in a small quantity, and most of this MBL would not take part in opsonization or complement activation.
It was reported very recently that a low concentration of MBL increases susceptibility to infection in patients with malignant disease, and thus the determination of the MBL genotype and serumprotein level would be useful for deciding treatment including replacement therapy 48–50. Our present method with FPLC and ELISA, which can simultaneously provide the level and genotype of MBL, would be a simple procedure for distinguishing the patients who are at high risk of infection. We also need to ascertain whether our findings are true for all sera, and further study should be carried out with a wide range of individuals including those of other ethnic groups.
4 Materials and methods
Serum levels of MBL were measured in 249 Japanese individuals whose MBL genotype was determined as previously reported 14, 51. Samples of peripheral blood were taken after obtaining informed consent and were used for MBL genotyping and determination of serum MBL concentration.
4.2 Gel filtration column chromatography
Freshly frozen serum from these individuals was diluted with an equal volume of elution buffer and centrifuged (13,000 rpm, 10 min). Gel filtration was carried out by FPLCTM (AmershamPharmacia Biotech, Uppsala, Sweden) using a 10 mm×30 cm SuperoseTM 6 column. Elutions were performed using a buffer referred to as buffer N, consisting of 20 mM Tris-HCl pH 7.4, 0.15 M NaCl, 0.05% NaN3, 0.05% Tween-20 and serine protease inhibitors, 100 μM p-amidinophenyl methanesulfonyl fluoride, 1 mM diisopropyl fluorophosphate and 100 μM p-nitrophenyl p′-guanidinobenzoate. Special care was taken to exclude Ca2+ or EDTA from buffer N in order to prevent Ca2+-dependent binding of MBL to the column resin (Superose consisting of agarose) as well as to prevent EDTA-induced dissociation of the MBL-MASP complex. A 100-μl aliquot of diluted serum was loaded on the FPLC column at a flow rate of 0.5 ml/min, and 0.5-ml fractions were collected. In the binding of MBL to Superose, elutions were performed at 4°C with buffer N with or without 1 mM CaCl2.
The column was calibrated with the following standard proteins: human IgM, 900,000 MW; human alpha-2 macroglobulin, 720,000 MW; bovine thyroglobulin, 669,000 MW; horse apoferritin, 440,000 MW;bovine catalase, 232,000 MW; rabbit aldolase, 158,000 MW and BSA, 69,000 MW.
4.3 ELISA procedure
4.3.1 MBL concentration
MBL levels were determined by a sandwich ELISA as described previously 26, 27. In brief, microtiter plates (MaxisorpTM; Nunc A/S, Roskilde, Denmark) were coated with a polyclonal anti-MBL rabbit IgG in PBSN (0.02 M PBS containing 0.05% NaN3). Purified serum MBL in a dilution buffer (PBS containing 0.01% thimerosal, 0.1% BSA and 0.1% Tween-20) was used as a standard. Samples diluted with this buffer were added to wells in duplicate and incubated at room temperature for 2 h. After washing with PBS, 100 μl of biotinylated anti-MBL mouse monoclonal IgG (3E7) 52, a kind gift of Dr. M. Matsushita, Fukushima Medical School, was added to each well and incubated overnight at 4°C. After washing, streptavidin-conjugated horseradish peroxidase (HRP) in the dilution buffer was added to the wells and left at room temperature for 90 min. After washing as before, wells were treated with 100 μl/well of o-phenylendiamine at 1 mg/ml in a substrate buffer (0.1 M citrate, 0.2 M Na2PO4 pH 5.5, 0.02% H2O2, 0.01% thimerosal). After a 30 min incubation at room temperature, the reaction was stopped by adding 100 μl/well of 2 N HCl, and the absorbance at 492 nm was read on a microplate reader.
4.3.2 Mannan binding assay
Binding of MBL to mannan was detected by the addition of samples to wells of a microtiter plate (Maxisorp) which were coated with 100 μl/well of mannan (from Saccharomyces cervisiae, Sigma Co., St. Louis, MO) at 100 μg/ml in carbonate-bicarbonate buffer (15 mM Na2CO3, 35 mM NaHCO3, pH 9.6) and subsequent reaction with the biotinylated anti-MBL antibody. In this binding experiment, 1 μl of 4 M CaCl2 was added to each well containing 100 μl of sample to a final concentration of 40 mM CaCl2, and incubation was carried out at 4°C overnight. Color development was performed in the same manner as in the MBL ELISA mentioned above. For the quantification of MBL which bound to mannan, values were calculated from calibration curves constructed with serially diluted serum from an individual with a high MBL concentration which was arbitrarily assigned mannan-binding MBL level of 1,000 U/l.
4.3.3 Detection of the MBL-MASP complex
Detection of the MBL-MASP complex was carried out by the addition of sample to the wells of the microplate (Maxisorp) which were coated with anti-MASP-1 monoclonal antibody 1E2 53, kindly provided by Dr. M. Matsushita, Fukushima Medical School, and subsequent reaction with the biotinylated anti-MBL antibody 3E7 40, 53. In this procedure, 75 μl of sample was added to wells of a microtiter plate, each containing 25 μl of buffer with Ca2+ (20 mM CaCl2, 20 mM Tris pH 7.4, 0.1% BSA, 0.01% thimerosal, 0.1% Tween-20, 0.15 M NaCl) and incubated overnight at 4°C. The procedure which followed was the same as that performed in the MBL ELISA mentioned above. Since the monoclonal anti-MASP-1 antibody (1E2) used in the present study was shown to recognize the A chain that is common to both MASP-1 and MASP-3, the MBL-MASP complex detected here was expressed as MBL-MASP-1/3. For the quantification of the MBL-MASP-1/3 complex, values were calculated from calibration curves constructed with serially diluted serum from an individual with a high MBL concentration which was arbitrarily assigned MBL-MASP-1/3 level of 1,000 U/l.
4.4 SDS-PAGE and Western blotting
SDS-PAGE was carried out in a 12.5% gel, and blotting was performed using polyvinylidene difluoride membranes (ImmobilonTM-P; Millipore, Bedford, MA). Blots were incubated with a biotinylated mouse monoclonal anti-MBL (3E7) antibody or a commercially available another one (clone 131–1, subclass IgG1; Statens Serum Institut, Copenhagen, Denmark) and developed with streptavidin-conjugated HRP (Nichirei Co., Tokyo, Japan) followed by enhanced chemiluminescence detection system (ECLTM; Amersham Pharmacia Biotech, Uppsala, Sweden). Two prestained SDS-PAGE standards, Low Range (Bio-Rad Laboratories, CA) and RainbowTM low range colored protein molecular mass markers (Amersham Pharmacia Biotech) were used simultaneously.
4.5 N-terminal amino acid sequencing of MBL
Fractions containing MBL were separated by SDS-PAGE (12.5% polyacrylamide gel) and electroblotted onto polyvinylidene difluoride membranes (Immobilon-P). A protein band at 32 kDa correspondingto MBL was visualized with Ponseau S, and was cut out and sequenced on an Applied Biosystems 477A/120A sequencer (Applied Biosystems, Foster City, CA).
We are deeply indebted to Dr. M. Matsushita from the Department of Biochemistry, Fukushima Medical College, Japan, for supplying a monoclonal anti-MBL IgG (3E7) and an anti-MASP-1/3 IgG (1E2) antibody. This research was supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan.