Biochimica et Biophysica Acta 1774 (2007) 335 – 344 www.elsevier.com/locate/bbapap
Posttranslational modifications in human plasma MBL and human recombinant MBL Pia Hønnerup Jensen a,b , Inga Laursen c , Finn Matthiesen b , Peter Højrup a,⁎ a
Department of Biochemistry and Molecular Biology, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmark b NatImmune A/S, Fruebjergvej 3, box 3, DK-2100 Copenhagen, Denmark c Statens Serum Institut, Artillerivej 5, DK-2300 Copenhagen, Denmark Received 21 September 2006; received in revised form 10 November 2006; accepted 21 December 2006 Available online 3 January 2007
Abstract Mannan-binding lectin (MBL) is a complex serum protein that plays an important role in innate immunity. In addition to assuming several different oligomeric forms, the polypeptide itself is highly heterogeneous. This heterogeneity is due to post-translational modifications, which help to stabilize the intact protein in its active conformation. For the first time, positions and occupation frequency of partial hydroxylations and partial glycosylations are reported in MBL. Hydroxylation and glycosylation patterns of both recombinant and plasma derived MBL were determined, using a combination of mass spectrometry on reduced MBL and on enzyme cleaved MBL. Variations in the degree of hydroxylation and glycosylation seem to be an indigenous characteristic of collectins. In addition to these already known modifications, a new post-translational modification was identified. Cys216 (and occasionally also Cys202) was modified in trace amounts to dehydroalanine, as detected by a 34 Da mass loss. This impairs the formation of a disulphide bond in the carbohydrate recognition domain. The dehydroalanine was identified in similar small amounts in both recombinant and plasma-derived MBL. © 2007 Elsevier B.V. All rights reserved. Keywords: Hydroxylysine; Hydroxyproline; Dehydroalanine from cysteine; Collectin; Mass spectrometry; Plasma protein
1. Introduction Mannan-binding lectin (MBL) is part of the collectin family. The collectins comprise surfactant protein A [1], surfactant protein D [2], conglutinin [3], CL-43 [4], liver collectin 1 [5] and CL-46 [6]. The name of the family arises from the proteins having a collagen-like region and a lectin domain. The sugar binding specificity of the individual collectins is determined by the lectin domain, which is stabilized by the formation of two conserved intra-polypeptide disulphide bonds [7]. In addition, all the collectins contain 2 or 3 cysteines in the N-terminal Abbreviations: MBL, mannan-binding lectin; pMBL, plasma derived human MBL; rhMBL, recombinant human MBL; Hyp, hydroxyproline; DHA, dehydroalanine; CRD, carbohydrate recognition domain; HCCA, α-cyano-4hydroxycinnamic acid; SA, sinapic acid; DTT, dithiothreitol; TFA, trifluoro acetic acid; MetO, oxidized methionine ⁎ Corresponding author. Tel.: +45 65502371; fax: +45 65502467. E-mail address:
[email protected] (P. Højrup). URL: http://www.protein.sdu.dk/ (P. Højrup). 1570-9639/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.bbapap.2006.12.008
region that participate in inter-polypeptide disulphide bonds, holding subunits and oligomers together [8]. Collectin subunits consist of 3 polypeptides wound together in a coil, via the collagen-like region. Like collagen, these collagen-like regions are heavily modified by hydroxylation of prolines and glycosylated hydroxylysines that help stabilize the collagen coil and ensure secretion of the protein [9–12]. Collagen consists of a characteristic Gly–Xaa–Yaa repeat where prolines and lysines in the Y-position are prone to hydroxylation. The hydroxylysines can further be O-glycosylated by the disaccharide galactose–glucose [13,14]. This is a two-step synthesis and occasionally a partial glycosylation (only a galactose) is observed [11]. MBL plays an important role in human innate immunity. It recognizes and binds specific sugar patterns on pathogen surfaces leading to clearance of these pathogens by two mechanisms, opsonization and lysis. MBL functions in two ways: (1) by acting as a direct opsonin and enhancing uptake of the pathogen by macrophages [15,16] or (2) by activating the
336
P.H. Jensen et al. / Biochimica et Biophysica Acta 1774 (2007) 335–344
complement system by its associated specific protease, MASP2, leading to phagocytosis through interaction with complement receptors or to the formation of the membrane attack complex, resulting in lysis of the pathogen [17]. MBL deficiency is quite common in the general human population. It is caused by three well-known structural MBL variants coding for non-functional MBL [18–20]. Furthermore, promoter polymorphisms also contribute to low plasma levels of functional MBL and by this to the number of MBL deficient individuals [21]. In otherwise immune competent persons this will have no impact on the general well-being [22]. However, for small children with an immature adaptive immune system [23,22] or for immunecompromised individuals (i.e. cancer patients receiving chemotherapy [24,25]) the consequences can be quite severe. Many serious diseases have been associated with MBL deficiency and the results of several studies have indicated that MBL replacement therapy might help selected patient groups (for a review see [26]). Plasma-derived MBL has been purified for this purpose, [27,28]; however, the yields are relatively low, and the source material might be insufficient if the indications for administering MBL hold [29]. An alternative product may be recombinant human MBL, but this approach has to fulfill several important criteria when compared to plasma-derived MBL. The heterogeneity of MBL is present at two levels, i.e. in the modification of the polypeptide itself as well as in the oligomerization of polypeptides, making it impossible to make a recombinant protein fully identical to the plasma-derived protein. However, it is important not to introduce new modifications. This study shows that recombinant human MBL (rhMBL) does fall within the “window of heterogeneity” of plasma derived MBL (pMBL), and therefore rhMBL could be a viable alternative for treating MBL-deficiency associated diseases. 2. Materials and methods 2.1. Materials α-cyano-4-hydroxycinnamic acid (HCCA) was from Aldrich (recrystallized in boiling acetonitrile). Sinapic acid (SA) was from Fluka. All water was obtained from a PURELABultra system (Elga) and all chemicals were of analytical grade unless stated otherwise.
2.2. Proteins Four different MBL preparations were applied for characterization studies. 2.2.1. Sample P1 (N065-73A) This pMBL batch was purified by NatImmune A/S, Copenhagen, from a pool of plasma originating from 8 healthy volunteers. pMBL was captured from the serum, in the presence of Ca2+, using a Glucosamine Sepharose 4FF column (GE Healthcare) and subsequently eluted with TBS containing EDTA. The retained product was concentrated on a Source 30Q (GE Healthcare), and subsequently stored frozen until testing. 2.2.2. Sample P2 (N213-12D) This pMBL batch was prepared from a Cohn fraction III-like paste harvested from a large pool of human plasma, originating from blood donations, by Statens Serum Institut, Denmark, as described elsewhere [27]. In brief, the starting material was affinity purified, in the presence of Ca2+, on a Sepharose CL-4B column, and MBL eluted by mannose. Subsequently solvent-detergent treatment
was performed, followed by an anion-exchange chromatography and a gel filtration step. The final liquid product containing about 0.3 mg MBL/ml was stabilized by addition of human serum albumin (5 mg/ml). Prior to structural investigations, the sample was initially concentrated by centrifugation at 10,000×g in a Microcon YM-100 tube (Amicon). Subsequently, pMBL was depleted of albumin using size-exclusion chromatography on Superose12 column (GE Healthcare). 2.2.3. Sample R1 (N027-56A) This rhMBL batch was produced by NatImmune A/S using a human embryonic kidney cell line HEK293. rhMBL was captured from the cell culture supernatant in the presence of Ca2+, using a Glucosamine Sepharose 4FF column (GE Healthcare) and subsequently eluted with TBS containing EDTA. The retained product was concentrated on a Source30Q column and desalted on a Sephacryl S-400 HR Column (all column materials from GE Healthcare). Finally, the product was stored frozen until testing. 2.2.4. Sample R2 (N177-21M) This rhMBL was produced for NatImmune A/S using the same procedure as for N027-56A, however with a DNA removal step included (using Benzonase, Merck Darmstadt), and with additional microfiltration and nanofiltration steps.
2.3. Western blot The protein samples were separated by SDS-PAGE under non-reducing conditions using 3–8% tris-acetate gels (Novex/Invitrogen). 20 ng or 2 ng of protein were loaded per lane, and electrophoresis performed for 75 min at 125 V. The gel was blotted onto a prewetted 450 nm polyvinylidene difluoride membrane (Invitrogen) using the XCell II blot module and a blotting time of 135 min at 25 V. The membrane was blocked for 60 min in a Tris blocking buffer (10 mM Tris, 140 mM NaCl, 0.1%v/v Tween-20, pH 8.0), and the biotin-labeled anti-MBL antibody (HYB131-01B, 1 mg/ml, Antibody Shop, Gentofte, Denmark) was applied in 2000× dilution in blocking buffer with incubation over night. The membrane was washed and incubated with streptavidinhorseradish peroxidase conjugate in blocking buffer (RPN1231V, GE Healthcare) for 60 min with gentle shaking. The protein bands were finally visualized using the SuperSignal West Pico substrate (Pierce) and 30 s exposure to chemiluminescense sensitive film (Hyperfilm ECL, GE Healthcare).
2.4. MALDI-MS analysis of MBL polypeptide chain and peptides MALDI MS analysis of MBL polypeptide chain was performed on a Voyager-DE™ STR (Applied Biosystems) controlled by the Data Explorer software (versions 3.4.0.0. and 4.4, Applied Biosystems). 0.5 μg of all four MBL preparations was freeze-dried, reconstituted in 10 mM dithiothreitol (DTT) and incubated for 10 min at 56 °C. Approximately 10 pmol was loaded in SA (20 μg/μl in 70% acetonitrile, 0.1% TFA) using the sandwich method (dried droplet [30] on top of thin layer [31]). Occasionally the samples were concentrated over a reverse phase micro-column (POROS® 50 R1, Perseptive Biosystems, 1-1259-06) packed in a GelLoader tip (Eppendorf) as previously described by Gobom et al. [32]. The sample was eluted onto the SA-coated target with a 20 μg/μl solution of SA (70% acetonitrile, 0.1% TFA). MALDI MS analysis of peptides was performed as described above. Alternatively, a 4700 Proteomics Analyzer from Applied Biosystems controlled by the Data Explorer software (version 4.4, Applied Biosystems) was used. Approximately 1 pmol of sample was loaded in HCCA (20 μg/μl in 70% acetonitrile, 0.1% TFA) using the dried droplet method [30]. HCCA 10 μg/μl was used for pMBL.
2.5. MALDI-MS/MS analysis MALDI-MS/MS analysis of peptides was performed on a 4700 Proteomics Analyzer from Applied Biosystems controlled by the Data Explorer software (version 4.4, Applied Biosystems). Approximately 10 pmol were concentrated over a reverse phase micro-column (POROS® 50 R2, Applied Biosystems, 1-1159-05) packed in a GelLoader tip (Eppendorf) as previously described by Gobom et al. [32]. The samples were eluted onto the target with HCCA (10 μg/μl in 70% acetonitrile, 0.1% TFA).
P.H. Jensen et al. / Biochimica et Biophysica Acta 1774 (2007) 335–344
337
2.6. Amino acid analysis Amino acid analyses were done by BioCentrum, Danish Technical University, Denmark. MBL was degraded into amino acids using acid hydrolysis [33]. Free amino acids were separated at 62 °C by cation exchange chromatography on a MCI-CK10U column (Mitsubishi Chemical Industries Ltd.), and eluted with a pH gradient. Immediately after separation, samples were oxidized with hypochlorite to oxidize proline and hydroxyproline to their corresponding primary amino acids. Subsequently, amino acids were derivatized with ortho-phthalaldehyde and 3-mercaptopropionic acid, and detected by fluorescence at 450 nm after excitation at 338 nm.
2.7. Proteolytic digestions MPG® Long Chain Alkylamine beads (CPG Inc.) were coated with 50 μg porcine trypsin as described elsewhere [34]. 100 μg of all four MBL preparations were freeze-dried and reconstituted in 50–60 μl trypsin-coated beads in 50 mM NH4HCO3, pH 7.6. The digestions were incubated 24 h at 37 °C while shaking at 700–1000 rpm. Digestions were terminated by removal of the magnetic beads. The digestions were kept at − 20 °C until RP-HPLC separation.
2.8. RP-HPLC All proteolytic digestions were separated using either the Ettan LC-900 (rhMBL) or the Äkta LC-900 (pMBL + rhMBL) system, both controlled by Unicorn software (version 3.21.02, Amersham Pharmacia Biotech). All separations were performed using a 2% gradient from 5 to 25% B solvent, 1.25% gradient from 25 to 50% and 3% gradient from 50 to 80% B solvent. Solvent A was 0.06% TFA and solvent B was 90% acetonitrile (80% for one of the rhMBL runs), in 0.05% TFA. The digestions of MBL were separated on a Jupiter 5μ C18 250 × 2.00 mm column. All native and reduced fractions were analyzed by MALDI MS.
2.9. Data analysis All data obtained on the Voyager-DE™ STR was processed using the m/z software (Version: 2001.08.14, Proteomics Inc.). Data obtained on the 4700 Proteomics Analyzer was analyzed using the Data Explorer software (version 4.4, Applied Biosystems). External calibration of MBL polypeptide spectra was performed using apomyoglobin (equine, Sigma) External calibration of peptide spectra was performed using tryptic peptides of lactoglobulin (bovine, Sigma). Peptides and modifications were identified using the GPMAW software (versions 5.03a and 6.2, Lighthouse data, Denmark) and the sequence of human MBL (Swiss Prot accession number P11226).
3. Results 3.1. Oligomeric distribution of pMBL and rhMBL Western blot analysis of a non-reduced and denatured, pMBL and rhMBL preparation shows similar, very complex, ladder-like oligomerization patterns (Fig. 1). The densest bands correspond to the oligomers of full subunits (3 polypeptides wound in a collagen-like coil). It appears that denatured rhMBL forms larger oligomers than denatured pMBL. The dominant forms of pMBL are trimers and tetramers of subunits, while the dominant forms of rhMBL are tetramers, pentamers and hexamers of subunits. Several other oligomers of intact subunits are visible as well as some intermediate forms, containing “unfinished” subunits. This indicates a very complex and diverse oligomerization process. The assignment of the individual oligomers in Fig. 1 is based on stepwise extrapolation from the small bands to the larger in a coomassie stained
Fig. 1. Western blot illustrating the oligomeric distribution in two preparations of denatured pMBL (P1 and P2) and two preparations of denatured rhMBL (R1 and R2). The dominant bands in P1 and P2 (pMBL) are the bands corresponding to trimers and tetramers of subunits, while the dominant bands in rhMBL are tetramers, pentamers and hexamers of subunits. In addition, several other oligomeric forms of MBL appear in both preparations.
gel. The size of the individual oligomers in the preparations were also verified by running a standard BioAnalyzer Protein200 assay (Agilent Technologies) under non-reducing conditions, including a standard protein mix ladder as reference. The trimer and tetramer bands run just above the 210 kDa myosin marker, while residual bands resulting from single polypeptide, polypeptide-dimer, subunit monomer and dimer run below the 210 kDa myosin marker with sizes consistent with the protein ladder (data not shown). 3.2. Characterization of the MBL polypeptide chain Upon reduction, all MBL oligomers fall apart and yield the MBL polypeptides. Also MALDI MS spectra of reduced pMBL and rhMBL reveal clear similarities (Fig. 2). A cluster of peaks ranging from MH+ 24815 to 25877 is seen in both preparations. Both clusters consist of two major peaks (25119 and 25448 for pMBL and 25149 and 25477 for rhMBL) along with several minor peaks. Slight variations in the mass values of the pMBL and the corresponding rhMBL peaks appear with the pMBL values being slightly lower. The clusters arise from the variation in glycosylation on hydroxylysines (Hyl) of the polypeptides, and the variation in the mass values between the two preparations is likely due to variation in the hydroxyproline (Hyp) content. Table 1 shows a summary of the measured mass values and the corresponding modifications. 3.3. Amino acid analysis Amino acid analysis was performed to verify and estimate the contents of hydroxylysine (Hyl) and hydroxyproline (Hyp)
338
P.H. Jensen et al. / Biochimica et Biophysica Acta 1774 (2007) 335–344
Hydroxyproline and hydroxylysine were also detected in pMBL (Sample P1), however, exact quantification of the amino acid composition was generally difficult, presumably caused by the presence of non-MBL related impurities (such as residual albumin and trace amounts of other proteins in Sample P2). 3.4. Separation and characterization of MBL peptides Tryptic digests of pMBL and rhMBL, respectively, were separated by reverse-phase HPLC (Fig. 3): peptides in all collected fractions were identified by MALDI MS and MALDI MS/MS. Table 2 presents a summary of all these peptides and the modifications identified. Fig. 4 illustrates the identified peptides and modifications in the intact MBL sequence. Some tryptic cleavage sites of MBL are always cleaved, while others are only partially cleaved, mainly due to posttranslational modifications. This gives rise to families of peptides that in the following run-through of the results are referred to as “segments”. 3.5. Segment 1: Glu1 to Arg32 Starting from the N-terminal, segment 1 (Glu1–Arg32) can be cleaved in several pieces. The peptide Glu1–Lys29 contains Fig. 2. MALDI MS spectra of reduced pMBL (sample P1) and rhMBL (Sample R1). MALDI MS of reduced MBL yields a cluster of peaks varying from MH+ 24815 to 25877. The peak cluster reflects the variation in glycosylation of hydroxylysines and hydroxylation of prolines. The major peaks in both preparations (MH+ 25119 and 25448 in pMBL and MH+ 25149 and 25477 in rhMBL) correspond to MBL polypeptide being glycosylated on three and four hydroxylysines, respectively. The rhMBL MH+ values are a little higher than the pMBL values, reflecting a higher degree of hydroxylation of prolines. The peaks with MH+ values higher than the 25448/25477 peaks, arise from sinapic acid adducts.
in rhMBL. In rhMBL (sample R2), 6.2 Hyp-residues and 12.0 Pro-residues were detected per polypeptide chain, corresponding to the 18 proline residues to be expected from the gene sequence. Similarly, 3.7 Hyl-residues and 14.9 Lys-residues were detected per polypeptide chain, corresponding to the 19 lysine residues to be expected from gene sequence. Similar results were obtained from Sample R1 (data not shown).
Table 1 MALDI MS peak clusters of pMBL and rhMBL Sample P1 (pMBL)
Sample R1 (rhMBL)
MH+
Hyl–Gal–Glc
24,815 ± 25 24,973 ± 25 25,119 ± 25 25,320 ± 25 25,448 ± 25 25,648 ± 26
2 7–8 2 1/2 6–7 3 5–6 3 1/2 6–7 4 4–5 Sinapic acid adduct
Hyp
MH+
Hyl–Gal–Glc
24,821 ± 25 25,017 ± 25 25,149 ± 25 25,320 ± 25 25,477 ± 26 25,684 ± 26 25,877 ± 26
2 7–8 2 1/2 8 3 7–8 3 1/2 6–7 4 6–7 Sinapic acid adduct Sinapic acid adduct
Hyp
MALDI MS of reduced MBL yields a cluster of peaks illustrating the variation in glycosylation of hydroxylysines (Hyl–Gal–Glc) and hydroxylation of prolines (Hyp) in the polypeptide. In addition some sinapic acid adducts to the polypeptide are seen. Also see Fig. 2.
Fig. 3. HPLC chromatograms of tryptic digests of Sample P1 (pMBL) and Sample R1 (rhMBL). Tryptic digests of pMBL (A) and rhMBL (B) were separated by reverse-phase HPLC and all fractions were collected for further analysis. Selected fraction numbers are indicated on the figure. The gradient is identical for both runs, but the B-solvent was 80% acetonitrile, 0.1% TFA in the rhMBL run, and 90% acetonitrile, 0.1% TFA in the pMBL run.
P.H. Jensen et al. / Biochimica et Biophysica Acta 1774 (2007) 335–344
339
Table 2 MALDI MS and MALDI MS/MS identification of peptides and modifications from tryptic digests of pMBL and rhMBL Peptide (aa)
Expected MH+
Observed MH+
1–10 1–29
1123.5 2923.4
11–29
1818.9
1123.6 2938.2 2955.2 1816.8 1832.6 1850.8 2146.9 2162.9 2178.9 1529.6 1544.6 1560.6 1576.6 1722.6 1738.6 1868.5 1884.7 1900.6 2046.4 2062.6 2208.8 2224.6 1467.9 1483.5 829.4 753.3 769.3 1447.7 1463.7 3828.4 3844.0 3859.1 3875.0 3991.3 4006.0 4021.0 4037.0 4136.3 4152.4 4167.0 4183.1 4199.0 2398.4 1291.3 1047.6 1063.6 919.5 935.5 1321.7 1079.5 951.4 1576.9 1592.4 3325.0 2812.6 2846.6
11–32
33–47
2147.0
1528.8
37–47
1127.6
40–47 48–55
813.4 753.4
56–71 56–92
1415.7 3455.7
72–92 80–92 93–101
2059.0 1291.6 1047.6
94–101
919.5
102–112 104–112 105–112 118–130
1320.8 1079.6 951.5 1576.8
(133–137)–(183–207) (133–137)–(208–228)
3326.6 2848.3
(133–137)–(208–228)– (183–207) 138–146 147–159 147–170
5641.6 976.5 1355.7 2588.3
5640.3 976.5 1356.8 2588.3
Modifications
pMBL
rhMBL
P1
P2
R1
R2
+ + +
+
Hyp27 Hyp21; Hyp27
+
+
+ + + + + + + + +
+
Hyp27 Hyp21; Hyp27
+
+ + + + + + + + + + + + + +
Hyp27 Hyp21; Hyp27 Hyl39 Hyl36; Hyl39 Hyl36; Hyl39; Hyp42 Hyl36; Hyl39–Gal Hyl36; Hyl39–Gal; Hyp42 Hyl39–Gal–Glc Hyl36; Hyl39–Gal–Glc Hyl36; Hyl39–Gal–Glc; Hyp42 Hyl36–Gal; Hyl39–Gal–Glc Hyl36–Gal; Hyl39–Gal–Glc; Hyp42 Hyl36–Gal–Glc; Hyl39–Gal–Glc Hyl36–Gal–Glc; Hyl39–Gal–Glc; Hyp42 Hyl39–Gal–Glc Hyl39–Gal–Glc; Hyp42 Hyp42 Hyp53 2 Hyp 3 Hyp Hyl71; Hyl74–Gal–Glc; 1 Hyp Hyl71; Hyl74–Gal–Glc; 2 Hyp Hyl71; Hyl74–Gal–Glc; 3 Hyp Hyl71; Hyl74–Gal–Glc; 4 Hyp Hyl71–Gal; Hyl74–Gal–Glc; 1 Hyp Hyl71–Gal; Hyl74–Gal–Glc; 2 Hyp Hyl71–Gal; Hyl74–Gal–Glc; 3 Hyp Hyl71–Gal; Hyl74–Gal–Glc; 4 Hyp Hyl71–Gal–Glc; Hyl74–Gal–Glc Hyl71–Gal–Glc; Hyl74–Gal–Glc; Hyp68 Hyl71–Gal–Glc; Hyl74–Gal–Glc; Hyp62; Hyp68 Hyl71–Gal–Glc; Hyl74–Gal–Glc; 3 Hyp Hyl71–Gal–Glc; Hyl74–Gal–Glc;
MetO99 MetO99
MetO126 Cys135–Cys202 Cys135–Cys224; DHA216 Cys135–Cys224 Cys135–Cys224/Cys216–Cys202
+ + + + + + + + + +
+ + + + + + + + + + + + + +
+ + + + + + + + + + + + + + + + + + + + +
+ + + + + + + + + + + + + + + + +
+ + + + + +
+ + + + + + +
+ + + + + + + + + + + + + + + + +
+ + + +
+ + + +
+ + + +
+ + + + + +
+ + + + + + + + + + + + + + + +
15 15–16 15–16 15 23 15 15 15 15 4 4 4 4–6 4 4–6 8–9 4–6 4–6 8–9 4–6 8–9 4–6 5 5 7 8 6 9 8 13 14 11 11 13 14 11 11 13 13 11 11 11 9 9–10 12 12 13 13 7 26 30–31 31–34 26–27 25 29–32 29–32
+ + + +
31–32 14 19–20 33–34
+
+ + + + + + + + + + +
+ + + +
+ +
+ +
HPLC fraction
+ + +
(continued on next page)
340
P.H. Jensen et al. / Biochimica et Biophysica Acta 1774 (2007) 335–344
Table 2 (continued) Peptide (aa)
Expected MH+
Observed MH+
147–182 160–170 160–182 171–182 183–207 183–207 (183–207)–(208–228)
3906.0 1251.6 2569.2 1336.7 2795.3
208–228 208–228
2317.0
3906.8 1251.1 2569.2 1336.8 2761.0 2795.3 5075.1 5108.5 2282.5 2317.0
5110.3
Modifications
pMBL
DHA202 Cys202–Cys224; DHA216 Cys202–Cys216 DHA216
rhMBL
P1
P2
R1
R2
+ + + + + +
+ + + +
+ + + + + + + + + +
+ + + + + +
+ + +
+
+ +
+ + +
HPLC fraction 35–37 21 25–26 17 27 29–32 36 31 29–32 30–32
Tryptic digests of pMBL and rhMBL were separated by reverse-phase HPLC and the fractions were analyzed by MALDI MS and MALDI MS/MS. Listed in the table are the expected MH+, the observed MH+, the modification identified and whether it is found in pMBL and/or rhMBL (+). Putative modification sites are the lysines Lys36, Lys39, Lys71 and Lys74 and the prolines Pro21, Pro27, Pro42, Pro53, Pro59, Pro62, Pro68 and Pro77. In addition MBL contains seven cysteines (Cys5, Cys12, Cys18, Cys135, Cys202, Cys216 and Cys224) that all participate in disulphide bonds. Abbreviations: Hyl = Hydroxylysine; Hyp = Hydroxyproline; DHA = dehydroalanine; MetO = oxidized methionine.
two putative hydroxylation sites (Hyp), at Pro21 and Pro27. In both pMBL and rhMBL a peptide containing Hyp27, as well as a peptide containing Hyp21 and Hyp27 were identified. The overlapping peptide Thr11–Lys29 is fully hydroxylated in pMBL, whereas the unmodified, partially modified and fully modified peptide was identified in rhMBL. Another overlapping peptide Thr11–Arg32 was only identified in rhMBL,
and shows the same modification profile. The specific assignment of hydroxylations at Pro21 and Pro27 is based on MS/MS of representative peptides. This segment however, was observed in several different HPLC fractions, indicating that different hydroxylation patterns may be present. The three cysteines in the N-terminal part of MBL participate in the oligomerization of the molecule. The disulphide bonding
Fig. 4. Sequence of MBL. The seven cysteines are marked in bold, the lysines prone to hydroxylation and glycosylation are underlined and the hydroxylated prolines for pMBL. The presence or absence are in italics. The sequence covered by the peptides identified in the two preparations is marked with — for rhMBL and of modifications is indicated: found (+) or not found (–), or detected both with and without (+/–). Major trypsin cleavage sites are indicated and give rise to the “peptide segments” mentioned in the text and illustrated in Table 2.
P.H. Jensen et al. / Biochimica et Biophysica Acta 1774 (2007) 335–344
patterns responsible for this oligomerization of rhMBL were solved as described elsewhere [35] and the same disulphide patterns were seen for pMBL (data not shown). 3.6. Segment 2: Asp33 to Arg47 Segment 2 (Asp33–Arg47) in the collagen-like region was also found as several sub-peptides. The peptide Asp33–Arg47 contains two putatively glycosylated hydroxylysines (Lys36 and Lys39) and one putative hydroxyproline (Pro42). In pMBL the peptide with no or only one hydroxylation was found, indicating partial hydroxylation of all three positions. In rhMBL at least two of the three sites were hydroxylated. In both rhMBL and pMBL, overlapping peptides Gly37–Arg47 and Gly40–Arg47 were found always to contain hydroxylated and glycosylated Lys39. Sometimes, also Pro42 was hydroxylated. The hydroxylysines were found as either partially glycosylated (only a galactose) or fully glycosylated (galactose–glucose). 3.7. Segment 3: Gly48 to Lys55 The next segment (Gly48–Lys55) was identified in all preparations, both with and without Hyp53. 3.8. Segment 4: Leu56 to Arg92 Segment 4 (Leu56–Arg92) also gives rise to several peptides. The peptide Leu56–Lys71 was always present; it was shown to contain two or three hydroxyprolines. The peptide Gly72–Lys92 contained a fully glycosylated hydroxylysine (Lys74) which was found in all preparations. Lys71 could be either hydroxylated, partially glycosylated or fully glycosylated. In the pMBL preparations, segment 4 was found to contain zero to four Hyps, while the frequency was slightly higher, two to four Hyps, in rhMBL. MALDI MS/MS was applied for identification of the particular Hyps in question: In a pMBL peptide containing one Hyp, the hydroxylation was located to Pro68. In an rhMBL peptide containing two Hyps, the hydroxylations were located to Pro62 and Pro68. A pMBL peptide containing three Hyps revealed hydroxylations at Pro59, Pro62 and Pro68, while an rhMBL peptide also containing three Hyps revealed occupation at Pro62, Pro68 and Pro77. 3.9. Segment 5: Lys93 to Ile228 (C-terminal) The C-terminal region segment 5 (Lys93–Ile228) contains the carbohydrate recognition domain (CRD). In both pMBL and rhMBL, all three cysteine-containing peptides (Ala133–Lys137, Leu183–Lys207 and Asn208–Ile228) were found to be interlinked by the conserved Cys135–Cys224 and Cys202–Cys216 bonds. In addition to the normal peptide, small amounts of the peptide Leu183–Lys207 were found to contain a modification, observed as a mass loss of 34 Da. The same 34 Da modification was found in the peptide Asn208–Ile228. MS/MS analysis identified the mass difference as due to the loss of SH2 from cysteine resulting in the formation of dehydroalanine (DHA) in
341
positions Cys202 and Cys216. The modifications were observed in trace amounts in both pMBL and rhMBL. Small amounts of the peptide Ala133–Lys137 were found to be disulphide linked to both a DHA-modified and a nonmodified peptide Asn208–Ile228, via the conserved Cys135– Cys224 bond. Similarly, the peptide Leu183–Lys207 was found to be disulphide linked to both a modified (only rhMBL) and a non-modified peptide Asn208–Ile228, via the conserved Cys202– Cys216 bond or via a Cys202–Cys224 bond (in the Cys216 modified peptide). Additionally, in both preparations trace amounts of the peptide Ala133–Lys137 were found to be disulphide linked to a non-modified Leu183–Lys207, via a Cys135–Cys202 disulphide bond. Additionally, this segment contains two methionines, which are both observed to be partially oxidized (MetO). This is a procedure induced modification. 3.10. Mapping coverage The total peptide-mapping experiments yield sequence coverage between 94.7% for pMBL and 97.0% for rhMBL (Fig. 4). All missing residues are situated in very small peptides with masses less than the detection limit (MH+ 700). This limit is set to prevent matrix signals from suppressing the peptide signals. 4. Discussion The heterogeneity of MBL can be observed at two levels: at the level of oligomerization where subunits of the same three polypeptides combine into various sized oligomers, and at the level of the individual polypeptide, as heterogeneity in the hydroxylation and glycosylation. In order for recombinant MBL to be a viable alternative as a therapeutic agent, it is essential that it does not introduce new modifications as potential immunogenic sites, but falls within the “window of heterogeneity” of the plasma-derived protein. Fig. 1 shows that the oligomerization pattern is similar in pMBL and rhMBL, although denatured rhMBL presents larger oligomeric forms than denatured pMBL. The dominant forms in pMBL are trimers and tetramers of the subunits as previously observed by Teillet et al. [36], while the dominant forms in rhMBL are tetramers, pentamers and hexamers of subunits. pMBL and rhMBL show similar oligomer structures in electron microscopy [35] and reveal highly similar retention times and peak profiles in size exclusion HPLC analysis (data not shown). Such observations indicate that the structures of pMBL and rhMBL are rather similar. That is, although the oligomeric distribution varies, with the higher oligomers dominating the recombinant preparation all species are present in both. Upon reduction, all MBL oligomers fall apart and yield MBL polypeptides, which on SDS-PAGE show an apparent mass of 32 kDa [37]. MALDI MS analysis of reduced MBL reveals an average mass of 25470 Da (Fig. 2), whereas the mass predicted from the sequence is 24014 Da without modifications. The larger apparent mass in SDS-PAGE is likely due to the
342
P.H. Jensen et al. / Biochimica et Biophysica Acta 1774 (2007) 335–344
abnormal mobility of collagen and collagen-like regions in SDS-PAGE. The collagens are rod-like, stretched out structures, which slow down the migration [38,39]. In MALDI MS analysis, pMBL and rhMBL again reveal large similarities (Fig. 2). Table 1 summarizes the peaks observed in Fig. 2 and states the modification to which these mass values most likely correspond. There is a clear increment of approximately 340 Da between the predominant peaks in each preparation (MH+ 25119 and 25448 in pMBL; MH+ 25149 and 25477 in rhMBL), corresponding to the disaccharide galactose–glucose (Gal–Glc, 340.1 Da) previously identified as the glycosylation moiety on the hydroxylysines in the collagenlike region [13,14]. These peaks correspond to three and four glycosylated lysines, respectively, out of four possible (Table 2; Fig. 4). The first peak in this series (MH+ 24815 and 24821) corresponds to two lysines being glycosylated. In between the major peaks are minor peaks with approximately 162 Da increments, corresponding to a lysine being modified with a galactose unit only [11]. The peaks with mass values higher than the fully glycosylated polypeptide (MH+ 25448 and 25477) arise from matrix adducts (sinapic acid, 224.2 Da) [40]. Table 1 also shows the approximate number of hydroxylated prolines (Hyp) in each species. The slightly higher mass values for rhMBL compared to the ones for pMBL indicate that rhMBL contains more hydroxyproline than pMBL (Table 2). The contents of hydroxylysine and hydroxyproline are confirmed by amino acid analysis. However, it is not possible to tell the positions of all hydroxyprolines from the combined data of the spectra of the intact polypeptide and the amino acid analysis. Therefore, tryptic digests of the two MBL preparations were performed in order to obtain more accurate information on the degree and position of hydroxylation of prolines. The tryptic digests of pMBL and rhMBL were separated by reverse-phase HPLC showing almost identical elution profiles (Fig. 3). Table 2 summarizes all the peptides and their modifications identified by MALDI MS and/or MALDI MS/MS analysis of the collected HPLC fractions, and Fig. 4 locates them to the sequence of MBL. The higher number of hydroxyprolines observed in the collagen-like region of rhMBL in Fig. 2 and confirmed by amino acid analysis is also observed at the peptide level. In segment 1 (Glu1–Arg32) hydroxyprolines were identified at positions 21 and 27 in all preparations. Both positions were also seen to be unmodified in rhMBL, whereas only Pro21 was unmodified in pMBL. From an overall view, only a few Nterminal peptides were identified in the analysis of pMBL compared to rhMBL, which may be an instrument issue. We have previously observed that the N-terminal peptides were difficult to isolate, due to low absorption at UV214 and to suppression effects in MALDI MS [35]. This indicates that the difference between pMBL and rhMBL in this region is likely due to lack of detection of the peptides. Segment 2 (Asp33–Arg47) and segment 3 (Gly48–Lys55) show an even distribution of hydroxyproline versus proline in pMBL and rhMBL. In segment 4 (Leu56–Arg92) all four
putative hydroxyprolines are identified in both pMBL and rhMBL, but the occupancy of the sites is higher in rhMBL, leading to a higher hydroxyproline content. MALDI MS/MS analysis of selected species identified the exact position of the hydroxyprolines. A pMBL peptide with one hydroxyproline, revealed Pro68 to be modified. An rhMBL peptide with two hydroxyprolines, revealed Pro62 and Pro68 to be modified. A pMBL peptide with three hydroxyprolines revealed Pro59, Pro62 and Pro68 to be modified, while a similar peptide from the rhMBL preparation revealed Pro62, Pro68 and Pro77 to be modified. We speculate that this indicates Pro68 to be modified first, then Pro62, and subsequently either Pro59 or Pro77, rather than this being interpreted as a difference between pMBL and rhMBL. In the N-terminal region a higher level of hydroxylation was observed in pMBL. However, only a few N-terminal peptides were identified as previously discussed. The hydroxyproline content of pMBL determined in this study is in accordance with our previous results [41,42] and the previous indications that rhMBL has a higher content of hydroxyproline [29] was confirmed. The glycosylation profiles observed at the polypeptide level (Fig. 2) and the peptide level (Table 2) yield complementary information. Cleavages in pMBL and rhMBL at Lys36, Lys39 and Lys71 (Table 2) indicate that these sites are not always glycosylated, whereas Lys74 always seems to be. Amino acid analysis verifies that all four lysines are hydroxylated, while peptide mapping indicates that they are not always glycosylated (non-glycosylated hydroxylysines can be cleaved by trypsin [43]). Fig. 2 indicates that the fully glycosylated form is the most abundant in both preparations and the tri-glycosylated form is more abundant in the pMBL preparation than in the rhMBL preparation. Previous studies have also shown the fully glycosylated MBL to be the most predominant form [41,42]. In the C-terminal part of the protein a previously undescribed modification of MBL was found. Although the major amount of the four cysteines was found to be disulphide linked Cys135–Cys224 and Cys202–Cys216 as previously described [7], a small amount of Cys202 and Cys216 was found to be modified to dehydroalanine (DHA). The amount of modified cysteine was <2%, as estimated from the relative absorption of the peptides based on an HPLC separation of reduced and alkylated tryptic MBL peptides (data not shown). When DHA was present, it occasionally resulted in the observation of two unusual disulphide bonds, Cys135–Cys202 and Cys202–Cys224 in trace amounts. We speculate that these bonds arise from scrambling of DHA modified peptides during digestion. In one preparation of rhMBL, the “free” cysteines were found to be modified by glutathione (peptide mass increase 305 Da, data not shown). The DHA modification has the consequence of inhibiting the formation of the Cys202–Cys216 disulphide bond. As the normal MBL polypeptide and the DHA variant only differ by the 34 Da lost in the formation of DHA from cysteine, it has not been possible to separate them. It has also not been possible to assign any function to the modification
P.H. Jensen et al. / Biochimica et Biophysica Acta 1774 (2007) 335–344
as the impact on the protein can only be studied in subunits or oligomers consisting of only the DHA variant. We speculate that the DHA variant is likely to result in changed sugar binding affinity, as the disulphides in the C-terminal region are responsible for stabilizing the carbohydrate recognition domain [7]. However, the content of DHA is very low, and the occasional presence of such a variant polypeptide in a large oligomer of MBL is not likely to have any effect on the final avidity of the complete oligomer. A search of the Swiss Prot protein database (The SRSWWW server at ExPASy, version 5.1.0) for DHA yielded a short list of entries containing only DHA originating from serine. The list contained a family of proteins: the ammonia lyases; and a peptide hormone family: the lantibiotics. The search yielded no records of DHA from cysteine modification. It is, however, a modification known to arise from alkaline treatment or high temperatures for long periods of time [44], and it can also be induced during collision-induced dissociation (CID) mass spectrometry [45,46]. However, this is not the case here, as (1) the protein is not subjected to alkaline conditions or high temperature treatment, (2) the peptide containing the modification is separated from the non-modified peptide prior to mass spectrometry analysis, and (3) the partnering cysteine can be modified by glutathione, indicating that it was in the reduced form prior to sample handling. rhMBL falls within the “window of heterogeneity” of pMBL as all modifications observed in rhMBL were also found in pMBL. The only detectable difference between the two MBL preparations was the variation in the content of Hyp and glycosylated Hyl in the spectra of the intact molecule. This leads us to propose that the higher content of Hyp and/or glycosylated Hyl in rhMBL is responsible for stabilizing the collagen coil and thereby strengthening the oligomerization process, leading to formation of higher oligomeric forms in rhMBL. The connection between Hyp content and degree of oligomerization has been suggested in previous studies [11,47]. Additionally, Teillet et al. [36] observe a slightly lower polypeptide mass, indicative of less hydroxyproline content, in their structural study of trimers and tetramers of MBL from human plasma. The hypothesis could be substantiated by fractionation of the oligomers of MBL, followed by determination of the average content of Hyp by amino acid analysis. Acknowledgements We thank NatImmune A/S and The Danish Ministry of Science, Technology and Innovation for financing. References [1] R.T. White, D. Damm, J. Miller, K. Spratt, J. Schilling, S. Hawgood, B. Benson, B. Cordell, Isolation and characterization of the human pulmonary surfactant apoprotein gene, Nature 317 (1985) 361–363. [2] A. Persson, D. Chang, K. Rust, M. Moxley, W. Longmore, E. Crouch, Purification and biochemical characterization of Cp4 (Sp-D) a collagenous surfactant-associated protein, Biochemistry 28 (1989) 6361–6367. [3] A.E. Davis, P.J. Lachmann, Bovine conglutinin is a collagen-like protein, Biochemistry 23 (1984) 2139–2144.
343
[4] U. Holmskov, B. Teisner, A.C. Willis, K.B. Reid, J.C. Jensenius, Purification and characterization of a bovine serum lectin (CL-43) with structural homology to conglutinin and SP-D and carbohydrate specificity similar to mannan-binding protein, J. Biol. Chem. 268 (1993) 10120–10125. [5] K. Ohtani, Y. Suzuki, S. Eda, T. Kawai, T. Kase, H. Yamazaki, T. Shimada, H. Keshi, Y. Sakai, A. Fukuoh, T. Sakamoto, N. Wakamiya, Molecular cloning of a novel human collectin from liver (CL-L1), J. Biol. Chem. 274 (1999) 13681–13689. [6] S. Hansen, D. Holm, V. Moeller, L. Vitved, C. Bendixen, K.B.M. Reid, K. Skjoedt, U. Holmskov, CL-46, a novel collectin highly expressed in bovine thymus and liver, J. Immunol. 169 (2002) 5726–5734. [7] W.I. Weis, K. Drickamer, W.A. Hendrickson, Structure of a C-type mannose-binding protein complexed with an oligosaccharide, Nature 360 (1992) 127–134. [8] S. Hansen, U. Holmskov, Structural aspects of collectins and receptors for collectins, Immunobiology 199 (1998) 165–189. [9] K.J. Colley, J.U. Baenziger, Posttranslational modifications of the corespecific lectin-relationship to assembly, ligand-binding, and secretion, J. Biol. Chem. 262 (1987) 10296–10303. [10] C.T. Heise, J.R. Nicholls, C.E. Leamy, R. Wallis, Impaired secretion of rat mannose-binding protein resulting from mutations in the collagen-like domain, J. Immunol. 165 (2000) 1403–1409. [11] Y. Ma, H. Shida, T. Kawasaki, Functional expression of human mannanbinding proteins (MBPs) in human hepatoma cell lines infected by recombinant vaccinia virus: post-translational modification, molecular assembly, and differentiation of serum and liver MBP, J. Biochem. (Tokyo) 122 (1997) 810–818. [12] C.A. Miles, A.J. Bailey, Thermally labile domains in the collagen molecule, Micron. 32 (2001) 325–332. [13] P. Bornstein, Biosynthesis of collagen, Ann. Rev. Biochem. 43 (1974) 567–603. [14] K.J. Colley, J.U. Baenziger, Identification of the posttranslational modifications of the core-specific lectin—The core-specific lectin contains hydroxyproline, hydroxylysine, and glucosylgalactosylhydroxylysine residues, J. Biol. Chem. 262 (1987) 10290–10295. [15] M. Kuhlman, K. Joiner, R.A. Ezekowitz, The human mannose-binding protein functions as an opsonin, J. Exp. Med. 169 (1989) 1733–1745. [16] M.W. Turner, Mannose-binding lectin: the pluripotent molecule of the innate immune system, Immunol. Today 17 (1996) 532–540. [17] M. Matsushita, T. Fujita, Activation of the classical complement pathway by mannose-binding protein in association with a novel C1s-like serine protease, J. Exp. Med. 176 (1992) 1497–1502. [18] P. Garred, S. Thiel, H.O. Madsen, L.P. Ryder, J.C. Jensenius, A. Svejgaard, Gene frequency and partial protein characterization of an allelic variant of mannan binding protein associated with low serum concentrations, Clin. Exp. Immunol. 90 (1992) 517–521. [19] R.J. Lipscombe, M. Sumiya, A.V. Hill, Y.L. Lau, R.J. Levinsky, J.A. Summerfield, M.W. Turner, High frequencies in African and non-African populations of independent mutations in the mannose binding protein gene, Hum. Mol. Genet. 1 (1992) 709–715. [20] H.O. Madsen, P. Garred, J.A. Kurtzhals, L.U. Lamm, L.P. Ryder, S. Thiel, A. Svejgaard, A new frequent allele is the missing link in the structural polymorphism of the human mannan-binding protein, Immunogenetics 40 (1994) 37–44. [21] H.O. Madsen, P. Garred, S. Thiel, J.A. Kurtzhals, L.U. Lamm, L.P. Ryder, A. Svejgaard, Interplay between promoter and structural gene variants control basal serum level of mannan-binding protein, J. Immunol. 155 (1995) 3013–3020. [22] M. Super, S. Thiel, J. Lu, R.J. Levinsky, M.W. Turner, Association of low levels of mannan-binding protein with a common defect of opsonisation, Lancet 2 (1989) 1236–1239. [23] J.A. Summerfield, M. Sumiya, M. Levin, M.W. Turner, Association of mutations in mannose binding protein gene with childhood infection in consecutive hospital series, BMJ 314 (1997) 1229–1232. [24] O. Neth, I. Hann, M.W. Turner, N.J. Klein, Deficiency of mannose-binding lectin and burden of infection in children with malignancy: a prospective study, Lancet 358 (2001) 614–618.
344
P.H. Jensen et al. / Biochimica et Biophysica Acta 1774 (2007) 335–344
[25] N.A. Peterslund, C. Koch, J.C. Jensenius, S. Thiel, Association between deficiency of mannose-binding lectin and severe infections after chemotherapy, Lancet 358 (2001) 637–638. [26] J.A. Summerfield, Clinical potential of mannose-binding lectin-replacement therapy, Biochem. Soc. Trans. 31 (2003) 770–773. [27] I. Laursen, Mannan-binding lectin (MBL) production from human plasma, Biochem. Soc. Trans. 31 (2003) 758–762. [28] H. Valdimarsson, M. Stefansson, T. Vikingsdottir, G.J. Arason, C. Koch, S. Thiel, J.C. Jensenius, Reconstitution of opsonizing activity by infusion of mannan-binding lectin (MBL) to MBL-deficient humans, Scand. J. Immunol. 48 (1998) 116–123. [29] J.C. Jensenius, P.H. Jensen, K. McGuire, J.L. Larsen, S. Thiel, Recombinant mannan-binding lectin (MBL) for therapy, Biochem. Soc. Trans. 31 (2003) 763–767. [30] M. Karas, F. Hillenkamp, Laser desorption ionization of proteins with molecular masses exceeding 10000 daltons, Anal. Chem. 60 (1988) 2299–2301. [31] O. Vorm, P. Roepstorff, M. Mann, Improved resolution and very highsensitivity in MALDI TOF of matrix surfaces made by fast evaporation, Anal. Chem. 66 (1994) 3281–3287. [32] J. Gobom, E. Nordhoff, E. Mirgorodskaya, R. Ekman, P. Roepstorff, Sample purification and preparation technique based on nano-scale reversed-phase columns for the sensitive analysis of complex peptide mixtures by matrix-assisted laser desorption/ionization mass spectrometry, J. Mass Spectrom. 34 (1999) 105–116. [33] V. Barkholt, A.L. Jensen, Amino-acid analysis—Determination of cysteine plus half-cystine in proteins after hydrochloric-acid hydrolysis with a disulfide compound as additive, Anal. Biochem. 177 (1989) 318–322. [34] T.N. Krogh, T. Berg, P. Hojrup, Protein analysis using enzymes immobilized to paramagnetic beads, Anal. Biochem. 274 (1999) 153–162. [35] P.H. Jensen, D. Weilguny, F. Matthiesen, K.A. McGuire, L. Shi, P. Hojrup, Characterization of the oligomer structure of recombinant human mannanbinding lectin, J. Biol. Chem. 280 (2005) 11043–11051. [36] F. Teillet, B. Dublet, J.P. Andrieu, C. Gaboriaud, G.J. Arlaud, N.M. Thielens, The two major oligomeric forms of human mannan-binding lectin: chemical characterization, carbohydrate-binding properties, and interaction with MBL-associated serine proteases, J. Immunol. 174 (2005) 2870–2877.
[37] J.H. Lu, S. Thiel, H. Wiedemann, R. Timpl, K.B. Reid, Binding of the pentamer/hexamer forms of mannan-binding protein to zymosan activates the proenzyme C1r2C1s2 complex, of the classical pathway of complement, without involvement of C1q. J. Immunol. 144 (1990) 2287–2294. [38] F. Larsen, H.O. Madsen, R.B. Sim, C. Koch, P. Garred, Disease-associated mutations in human mannose-binding lectin compromise oligomerization and activity of the final protein, J. Biol. Chem. 279 (2004) 21302–21311. [39] H. Furthmayr, R. Timpl, Characterization of collagen peptides by sodium dodecylsulfate-polyacrylamide electrophoresis, Anal. Biochem. 41 (1971) 510–516. [40] D. Suckau, A. Resemann, K. Köster, U. Rapp, A. Holle, Gridless delayed extraction MALDI-TOF spectra of larger proteins., Abstract from oral at The Australian and New Zealand Society of Mass Spectrometry Conference 16 (TuO-13) (1997) [41] U.G. Jensen, Purification and characterisation of the human lectin mannanbinding-protein, Master's Thesis, University of Southern Denmark, Odense Denmark (1996) 1–72. [42] P.H. Nielsen, Primary structure analysis of human MBL: disulphide bonds and other modifications, Master's Thesis, University of Southern Denmark, Odense Denmark (2001) 1–91. [43] S. Ruiz, A.H. Henschen-Edman, A.J. Tenner, Localization of the site on the complement component C1q required for the stimulation of neutrophil superoxide production, J. Biol. Chem. 270 (1995) 30627–30634. [44] M. Friedman, Chemistry, biochemistry, nutrition, and microbiology of lysinoalanine, lanthionine, and histidinoalanine in food and other proteins, J. Agric. Food Chem. 47 (1999) 1295–1319. [45] H. Steen, M. Mann, Similarity between condensed phase and gas phase chemistry: fragmentation of peptides containing oxidized cysteine residues and its implication for proteomics, J. Am. Soc. Mass Spectrom. 12 (2000) 228–232. [46] J. Yague, A. Nunez, M. Boix, M. Esteller, P. Alfonso, J.I. Casal, Oxidation of carboxyamidomethyl cysteine may add complexity to protein identification, Proteomics 5 (2005) 2761–2768. [47] T. Vorup-Jensen, E.S. Sørensen, U.B. Jensen, W. Schwaeble, T. Kawasaki, Y. Ma, K. Uemura, N. Wakamiya, Y. Suzuki, T.G. Jensen, K. Takahashi, A.B. Ezekowitz, S. Thiel, J.C. Jensenius, Recombinant expression of human mannan-binding lectin, Int. Immunopharmacol. (2001) 677–687.