Direct evidence that J chain regulates the polymeric structure of IgM in antibody-secreting B cells.

IgM is secreted in two functional polymeric forms. Secreted IgM was originally thought to be exclusively a pentameric molecule containing J (joining) chain, but many B cells also secrete hexameric IgM lacking J chain. Hexameric IgM may play an important role in the immune system, since it is up to 20 times more active than pentameric IgM in initiating the complement cascade. The predominant polymeric form of IgM secreted by B cell lines, either pentameric or hexameric, correlates with the concentration of J chain present during polymerization, and cells that express high levels of J chain secrete mostly IgM pentamers. The B cell lymphoma WEHI-231 does not express J chain, and the majority of its secreted IgM is polymerized as hexamers. When a J chain-encoding cDNA was expressed in these cells, the secreted IgM was found to be almost exclusively pentameric. However, although the expression of J chain dramatically altered the phenotype of the IgM secreted by these cells, it had little effect on their secretory rate. We conclude that J chain regulates the structure and function of the IgM polymers secreted by B cells, but it is not necessary for either IgM polymerization or secretion.     

Disulfide bond assignment in human J chain and its covalent pairing with immunoglobulin M.

The assignment of disulfide bonds in human J chain and its covalent pairing with immunoglobulin M was determined under conditions which minimize disulfide bond interchange. We show that in J chain the three intradisulfide bridges are formed between Cys 12 and 100, Cys 71 and 91, and Cys 108 and 133. Previous reports [reviewed by Koshland, M. E. (1985) Annu. Rev. Immunol. 3, 425-453] have proposed that cysteines 12, 14, or 68 were linked to the penultimate cysteine 575 of two mu chain tails. In this work, we demonstrate that cysteines 14 and 68 are disulfide-bridged to mu chains. A revised, albeit putative, model of J chain folding is presented which takes into account the correct disulfide pairing and the predictive secondary structure assignment.     

Polymerization of IgA and IgM: roles of Cys309/Cys414 and the secretory tailpiece

We have investigated how the secretory tailpiece (tp), Cys414 and the amino acids flanking Cys414 or Cys309 are involved in regulating the different polymerization of IgM and IgA to pentamers and dimers/monomers, respectively. Whereas changing the tp of IgM to that of IgA has little effect on IgM polymerization, introducing the mu tp to IgA leads to the formation of larger than wild-type IgA polymers, including pentamers and hexamer. This shows that the secretory tp can differentially regulate polymerization depending on the heavy chain context. Cys414, which is engaged in intermonomeric disulfide bonds in IgM, is not crucial for the difference in IgM and IgA polymerization; IgM with a C414S mutation forms more large polymers than IgA. Also, IgA with IgM-like mutations in the five amino acids flanking Cys309, which is homologous to Cys414, oligomerize similarly as IgA wild type. Thus, IgA appears to have an inherent tendency to form monomers and dimers that is partially regulated by the tp, while the Cys309 region has only a minor effect. We also show that complement activation by IgM is sensitive to alterations in the polymeric structure, while IgA is inactive in classical complement activation even for polymers such as pentamers and hexamers.     

Quality control of ER synthesized proteins: an exposed thiol group as a three-way switch mediating assembly, retention and degradation.

Plasma cells secrete IgM only in the polymeric form: the C-terminal cysteine of the mu heavy chain (Cys575) is responsible for both intracellular retention and assembly of IgM subunits. Polymerization is not quantitative, and part of IgM is degraded intracellularly. Neither chloroquine nor brefeldin A (BFA) inhibits degradation, suggesting that this process occurs in a pre-Golgi compartment. Degradation of IgM assembly intermediates requires Cys575: the monomeric IgMala575 mutant is stable also when endoplasmic reticulum (ER) to Golgi transport is blocked by BFA. Addition of the 20 C-terminal residues of mu to the lysosomal protease cathepsin D is sufficient to induce pre-Golgi retention and degradation of the chimeric protein: the small amounts of molecules which exit from the ER are mostly covalent dimers. By contrast, when retained by the KDEL sequence, cathepsin D is stable in the ER, indicating that retention is not sufficient to cause degradation. Replacing the C-terminal cysteine with serine restores transport through the Golgi. As all chimeric cathepsin D constructs display comparable protease activity in vitro, their different fates are not determined by gross alterations in folding. Thus, also out of its normal context, the mu chain Cys575 plays a crucial role in quality control, mediating assembly, retention and degradation.     

Mutations of the mouse mu H chain which prevent polymer assembly

Earlier work has shown that truncated mu-chains lacking the carboxy-terminal C mu 4-tail region are secreted as monomeric rather than polymeric IgM and that the monomer phenotype is not due to the lack of a disulfide bond at Cys-575 in the tail. In order to define with greater precision, the molecular requirements for IgM polymer assembly, we have isolated several mutant hybridomas which produce monomeric IgM. For three such mutants, we synthesized cDNA clones of their mu mRNA and identified a mutation in the mu-chain which was responsible for the failure to assemble polymers. Mutant 205 has a 2-bp deletion which results in a termination codon after amino acid 556, effectively deleting the last 20 amino acids of the mu-chain. In conjunction with earlier reports, this result shows that the tail plays some role in assembly other than providing Cys-575, the penultimate amino acid, for disulfide bond formation. Both mutant 21 and mutant 201 have an A to G transition, which results in Tyr-455 in the fourth constant domain being replaced by a cysteine. We conclude that the integrity of both the C mu 4 domain and the 19 amino acid tail are required for the mu H chain to be assembled into polymeric IgM.     

Primary structure of human alpha 2-macroglobulin. V. The complete structure

The primary structure of the tetrameric plasma glycoprotein human alpha 2-macroglobulin has been determined. The identical subunits contain 1451 amino acid residues. Glucosamine-based oligosaccharide groups are attached to asparagine residues 32, 47, 224, 373, 387, 846, 968, and 1401. Eleven intrachain disulfide bridges have been placed (Cys25-Cys63, Cys228-Cys276, Cys246-Cys264, Cys255-Cys408, Cys572-Cys748, Cys619-Cys666, Cys798-Cys826, Cys824-Cys860, Cys898-Cys1298, Cys1056-Cys1104, and Cys1329-Cys1444). Cys-447 probably forms an interchain bridge with Cys-447 from another subunit. The beta-SH group of Cys-949 is thiol esterified to the gamma-carbonyl group of Glx-952, thus forming an activatable reactive site which can mediate covalent binding of nucleophiles. A putative transglutaminase cross-linking site is constituted by Gln-670 and Gln-671. The primary sites of proteolytic cleavage in the activation cleavage area (the "bait" region) are located in the sequence: -Arg681-Val-Gly-Phe-Tyr-Glu-. The molecular weight of the unmodified alpha 2-macroglobulin subunit is 160,837 and approximately 179,000, including the carbohydrate groups. The presence of possible internal homologies within the alpha 2-macroglobulin subunit is discussed. A comparison of stretches of sequences from alpha 2-macroglobulin with partial sequence data for complement components C3 and C4 indicates that these proteins are evolutionary related. The properties of alpha 2-macroglobulin are discussed within the context of proteolytically regulated systems with particular reference to the complement components C3 and C4.     

Respiratory chain strongly oxidizes the CXXC motif of DsbB in the Escherichia coli disulfide bond formation pathway

Escherichia coli DsbB has four essential cysteine residues, among which Cys41 and Cys44 form a CXXC redox active site motif and the Cys104-Cys130 disulfide bond oxidizes the active site cysteines of DsbA, the disulfide bond formation factor in the periplasm. Functional respiratory chain is required for the cell to keep DsbA oxidized. In this study, we characterized the roles of essential cysteines of DsbB in the coupling with the respiratory chain. Cys104 was found to form the inactive complex with DsbA under respiration-defective conditions. While DsbB, under normal aerobic conditions, is in the oxidized state, having two intramolecular disulfide bonds, oxidation of Cys104 and Cys130 requires the presence of Cys41-Cys44. Remarkably, the Cys41-Cys44 disulfide bond is refractory to reduction by a high concentration of dithiothreitol, unless the membrane is solubilized with a detergent. This reductant resistance requires both the respiratory function and oxygen, since Cys41-Cys44 became sensitive to the reducing agent when membrane was prepared from quinone- or heme-depleted cells or when a membrane sample was deaerated. Thus, the Cys41-Val-Leu-Cys44 motif of DsbB is kept both strongly oxidized and strongly oxidizing when DsbB is integrated into the membrane with the normal set of respiratory components.     

Formation of intermolecular disulfide bonds on nascent immunoglobulin polypeptides.

The initial step of intermolecular covalent assembly of immunoglobulins molecules involves formation of heavy chain-light chain or heavy chain-heavy chain disulfide bonds. Using QAE-Sephadex chromatography to isolate microsomal nascent polypeptides, we have shown that this initial step of intermolecular covalent assembly occurs, to a substantial extent, on nascent heavy chains, as well as on completed heavy chains as previously demonstrated by others. In MPC 11 mouse myeloma cells, completed light chains are assembled covalently to nascent heavy chains, whereas in MOPC 21 mouse myeloma cells, completed heavy chains are assembled covalently to nascent heavy chains. These results are consisted with the heavy-light half-molecule being the major initial intermediate in the assembly of MPC 11 IgG2b and heavy-heavy dimer being the major initial intermediate formed in assembly of MOPC 21 IgG1. The nascent MPC 11 heavy chain must be at least 38,000 daltons in size before assembly with the light chain occurs, even though the heavy chain cysteine involved in this disulfide bond is 131 residues (approximately 15,000 daltons) from the NH2 terminus. In addition, pulse-chase labeling studies of MPC 11 cells have shown that the assembly of completed light chains with the nascent heavy chain must occur within a few minutes of the synthesis of the light chain even though a large excess of unassembled MPC 11 light chains remain inside the cell for an average time of 2 h before being secreted.     

Structural requirements for IgM assembly and cytolytic activity. Effects of mutations in the oligosaccharide acceptor site at Asn402

Glycosylation of IgG occurs at asparagine 297 of the gamma H chain and is necessary for the normal capacity of IgG to activate the classical pathway of complement-dependent cytolysis. IgM is glycosylated at five sites in the constant region of the mu H chain, of which glycosylation at asparagine 402 seems analogous to the glycosylation of IgG. In order to assess the importance of glycosylation at asparagine 402 for IgM cytolytic activity, we have used site-directed mutagenesis to produce IgM which is not glycosylated at this position. In particular we have tested the effects of substituting Gln for Asn 402 and Thr-Gly for Gly 403-Thr 404 in the third constant region domain. We tested the effects of these substitutions by expressing the mutant mu genes in hybridoma cells which produce the hapten-specific kappa-chain. The normal mu-chain is glycosylated at Asn 402, and, as expected, these mutations appear to abrogate glycosylation of the mutant mu-chains at position 402 and do not affect the hapten affinity of the IgM. However, both of these mutations cause the increased production of monomeric rather than polymeric IgM: the ratio of monomeric to polymeric IgM is 0.21, 3.5, and 10.3 for wild-type IgM, IgM-Gln 402, and IgM-Thr 403-Gly 404, respectively. The wild-type and mutant polymeric IgM preparations were compared for their capacity to promote complement-dependent cytolysis: IgM-Gln 402 and IgM-Thr 403-Gly 404 have approximately 31% and 4%, respectively, of the capacity of wild-type IgM.     

Identification of the disulfide bonds in human plasma protein SP-40,40 (apolipoprotein-J).

SP-40,40, a human plasma protein, is a modulator of the membrane attack complex formation of the complement system as well as a subcomponent of high-density lipoproteins. In the present study, the positions of the disulfide bonds in SP-40,40 were determined. SP-40,40 was purified from human seminal plasma by affinity chromatography using an anti-SP-40,40 monoclonal antibody and reversed-phase, high-performance liquid chromatography (HPLC). The protein was digested with trypsin and the fragments were separated by reversed-phase HPLC. The peptides containing disulfide bonds were fluorophotometrically detected with 4-(aminosulfonyl)-7-fluoro-2,1,3-benzoxadiazole (ABD-F). The peptides containing more than two disulfide bonds were further digested with Staphylococcus aureus V8 protease and lysylendopeptidase, and the fragments were isolated by HPLC. The amino acid compositions and the amino acid sequences of the peptides containing only a disulfide bond were determined. Disulfide bonds thus determined were between Cys58(alpha)-Cys107(beta), Cys68(alpha)-Cys99(beta), Cys75(alpha)-Cys94(beta), and Cys86(alpha)-Cys80(beta). Since there was no free sulfhydryl groups in the SP-40,40 molecule, Cys78(alpha) and Cys91(beta) should also be linked by a disulfide bond. It is notable that all of the disulfide bonds in SP-40,40 are not only formed by inter-chain pairing, but also appear to form an antiparallel ladder-like structure between the two chains. The unique structure could be related to the functions of SP-40,40.     

The disulfide bridges of the trypsin-kallikrein inhibitor K from snails (Helix pomatia). Thermal inactivation and proteolysis by thermolysin (author's transl)]

Isoinhibitor K is the main component of the complex mixture of isoinhibitors of broad specificity secreted into the mucus by the Roman snail (Helix pomatia). The disulfide pairing was determined after the amino acid sequence had been elucidated. Two cystine-containing peptides with the disulfide bridges Cys32-Cys53 and Cys32-Cys53 plus Cys7-Cys57 were obtained after thermolytic hydrolysis of the native inhibitor at 80 degrees C and chromatographic separation of the peptides using SE-Sephadex. The Cys16-Cys40 disulfide bridge could be reduced selectively by sodium borohydride with no loss in biological activity. This property and the covalent structure correspond to that of the intracellular inhibitor from bovine organs, which is largely homologous in its amino acid sequence to the secretory inhibitor from the snail. The complete covalent structure of isoinhibitor K will be presented. The snail inhibitor is less stable against proteolytic inactivation by thermolysin and against thermal denaturation at pH 8.0 than the inhibitor from bovine organs (Kunitz inhibitor).     

Structural requirements for polymeric immunoglobulin assembly and association with J chain

Both IgM and IgA exist as polymeric immunoglobulins. IgM is assembled into pentamers with J chain and hexamers lacking J chain. In contrast, polymeric IgA exists mostly as dimers with J chain. Both IgM and IgA possess an 18-amino acid extension of the C terminus (the tail-piece (tp)) that participates in polymerization through a penultimate cysteine residue. The IgM (mutp) and IgA (alphatp) tail-pieces differ at seven amino acid positions. However, the tail-pieces by themselves do not determine the extent of polymerization. We now show that the restriction of polymerization to dimers requires both C(alpha)3 and alphatp and that more efficient dimer assembly occurs when C(alpha)2 is also present; the dimers contain J chain. Formation of pentamers containing J chain requires C(mu)3, C(mu)4, and the mutp. IgM-alphatp is present mainly as hexamers lacking J chain, and mumugammamu-utp forms tetramers and hexamers lacking J chain, whereas IgA-mutp is present as high order polymers containing J chain. In addition, there is heterogeneous processing of the N-linked carbohydrate on IgA-mutp, with some remaining in the high mannose state. These data suggest that in addition to the tail-piece, structural motifs in the constant region domains are critical for polymer assembly and J chain incorporation.     

Isolation and characterization of disulfide-bonded peptides from the three globular domains of aggregating cartilage proteoglycan

The aggregating cartilage proteoglycan core protein contains two globular domains near the N terminus (G1 and G2) and one near the C terminus (G3). The G1-G3 domains contain 10, 8, and 10 cysteine residues, respectively. The disulfide assignments of the G1 domain have previously been deduced (Neame, P. J., Christner, J. E., and Baker, J. R. (1987) J. Biol. Chem. 262, 17768-17778) as Cys1-Cys2, Cys3-Cys6, Cys4-Cys5, Cys7-Cys10, and Cys8-Cys9, in which the numbers cited after the half-cystine residues are their relative positions from the N terminus. Here we describe a method for the isolation of disulfide-bonded peptides from tryptic digests of bovine nasal cartilage monomer. Sequence analysis of these peptides has allowed us to confirm the pairings previously determined for the G1 domain and to assign a disulfide pattern for the G2 domain of Cys11-Cys14, Cys12-Cys13, Cys15-Cys18, and Cys16-Cys17, in which the Cys15-Cys18 pairing was deduced indirectly. Similarly, for the G3 domain, a pattern of Cys19-Cys20, Cys21-Cys24, Cys22-Cys23, Cys25-Cys27, and Cys26-Cys28 was assigned, in which the Cys22-Cys23 pair was deduced indirectly. The G2 domain therefore contains disulfide bonding which is characteristic of the tandem repeat structures found in the G1 domain and link protein, and the G3 domain contains the three disulfide linkages previously assigned to the family of C-type animal lectins. The method described here, which combines anion-exchange, cation-exchange, and reversed-phase chromatography, should have broad application to the isolation of disulfide-bonded peptides from other heavily glycosylated proteins and proteoglycans.     

Localization of disulfide bonds in the cystine knot domain of human von Willebrand factor

von Willebrand factor (VWF) is a multimeric glycoprotein that is required for normal hemostasis. After translocation into the endoplasmic reticulum, proVWF subunits dimerize through disulfide bonds between their C-terminal cystine knot-like (CK) domains. CK domains are characterized by six conserved cysteines. Disulfide bonds between cysteines 2 and 5 and between cysteines 3 and 6 define a ring that is penetrated by a disulfide bond between cysteines 1 and 4. Dimerization often is mediated by additional cysteines that differ among CK domain subfamilies. When expressed in a baculovirus system, recombinant VWF CK domains (residues 1957-2050) were secreted as dimers that were converted to monomers by selective reduction and alkylation of three unconserved cysteine residues: Cys(2008), Cys(2010), and Cys(2048). By partial reduction and alkylation, chemical and proteolytic digestion, mass spectrometry, and amino acid sequencing, the remaining intrachain disulfide bonds were characterized: Cys(1961)-Cys(2011) (), Cys(1987)-Cys(2041) (), Cys(1991)-Cys(2043) (), and Cys(1976)-Cys(2025). The mutation C2008A or C2010A prevented dimerization, whereas the mutation C2048A did not. Symmetry considerations and molecular modeling based on the structure of transforming growth factor-beta suggest that one or three of residues Cys(2008), Cys(2010), and Cys(2048) in each subunit mediate the covalent dimerization of proVWF.     

Contribution of disulfide bonds to the conformational stability and catalytic activity of ribonuclease A.

Disulfide bonds between the side chains of cysteine residues are the only common crosslinks in proteins. Bovine pancreatic ribonuclease A (RNase A) is a 124-residue enzyme that contains four interweaving disulfide bonds (Cys26-Cys84, Cys40-Cys95, Cys58-Cys110, and Cys65-Cys72) and catalyzes the cleavage of RNA. The contribution of each disulfide bond to the conformational stability and catalytic activity of RNase A has been determined by using variants in which each cystine is replaced independently with a pair of alanine residues. Thermal unfolding experiments monitored by ultraviolet spectroscopy and differential scanning calorimetry reveal that wild-type RNase A and each disulfide variant unfold in a two-state process and that each disulfide bond contributes substantially to conformational stability. The two terminal disulfide bonds in the amino-acid sequence (Cys26-Cys84 and Cys58-Cys110) enhance stability more than do the two embedded ones (Cys40-Cys95 and Cys65-Cys72). Removing either one of the terminal disulfide bonds liberates a similar number of residues and has a similar effect on conformational stability, decreasing the midpoint of the thermal transition by almost 40 degrees C. The disulfide variants catalyze the cleavage of poly(cytidylic acid) with values of kcat/Km that are 2- to 40-fold less than that of wild-type RNase A. The two embedded disulfide bonds, which are least important to conformational stability, are most important to catalytic activity. These embedded disulfide bonds likely contribute to the proper alignment of residues (such as Lys41 and Lys66) that are necessary for efficient catalysis of RNA cleavage.     

Disulfide bonding arrangements in active forms of the somatomedin B domain of human vitronectin

The N-terminal cysteine-rich somatomedin B (SMB) domain (residues 1-44) of the human glycoprotein vitronectin contains the high-affinity binding sites for plasminogen activator inhibitor-1 (PAI-1) and the urokinase receptor (uPAR). We previously showed that the eight cysteine residues of recombinant SMB (rSMB) are organized into four disulfide bonds in a linear uncrossed pattern (Cys(5)-Cys(9), Cys(19)-Cys(21), Cys(25)-Cys(31), and Cys(32)-Cys(39)). In the present study, we use an alternative method to show that this disulfide bond arrangement remains a major preferred one in solution, and we determine the solution structure of the domain using NMR analysis. The solution structure shows that the four disulfide bonds are tightly packed in the center of the domain, replacing the traditional hydrophobic core expected for a globular protein. The few noncysteine hydrophobic side chains form a cluster on the outside of the domain, providing a distinctive binding surface for the physiological partners PAI-1 and uPAR. The hydrophobic surface consists mainly of side chains from the loop formed by the Cys(25)-Cys(31) disulfide bond, and is surrounded by conserved acidic and basic side chains, which are likely to contribute to the specificity of the intermolecular interactions of this domain. Interestingly, the overall fold of the molecule is compatible with several arrangements of the disulfide bonds. A number of different disulfide bond arrangements were able to satisfy the NMR restraints, and an extensive series of conformational energy calculations performed in explicit solvent confirmed that several disulfide bond arrangements have comparable stabilization energies. An experimental demonstration of the presence of alternative disulfide conformations in active rSMB is provided by the behavior of a mutant in which Asn(14) is replaced by Met. This mutant has the same PAI-1 binding activity as rVN1-51, but its fragmentation pattern following cyanogen bromide treatment is incompatible with the linear uncrossed disulfide arrangement. These results suggest that active forms of the SMB domain may have a number of allowed disulfide bond arrangements as long as the Cys(25)-Cys(31) disulfide bond is preserved.     

The carboxyl-terminal domains of IgA and IgM direct isotype-specific polymerization and interaction with the polymeric immunoglobulin receptor

Mucosal surfaces are protected by polymeric immunoglobulins that are transported across the epithelium by the polymeric immunoglobulin receptor (pIgR). Only polymeric IgA and IgM containing a small polypeptide called the "joining" (J) chain can bind to the pIgR. J chain-positive IgA consists of dimers, and some larger polymers, whereas only IgM pentamers incorporate the J chain. We made domain swap chimeras between human IgA1 and IgM and found that the COOH-terminal domains of the heavy chains (Calpha3 and Cmu4, respectively) dictated the size of the polymers formed and also which polymers incorporated the J chain. We also showed that chimeric IgM molecules engineered to contain Calpha3 were able to bind the rabbit pIgR. Since the rabbit pIgR normally does not bind IgM, these results suggest that the COOH-terminal domain of the polymeric immunoglobulins is primarily responsible for interaction with the pIgR. Finally, we made a novel chimeric IgA immunoglobulin, containing the terminal domain from IgM. This recombinant molecule formed J chain-containing pentamers that could, like IgA, efficiently form covalent complexes with the human pIgR ectodomain, known as secretory component.     

Mutations affecting the structure and function of immunoglobulin M.

Using a hybridoma cell line which secretes hapten-specific immunoglobulin M (IgM), we have isolated a variety of mutants which produce abnormal immunoglobulin. Immunoglobulin was tested for the size and composition of the component heavy and light chains and for variable and constant region related functional and serological activities. Some mutants secrete IgM which seems to be defective in hapten binding; others make IgM which appears not to activate complement. Many of the mutants secrete monomeric as opposed to pentameric IgM. In some cases, the defect apparently correlates with structural alterations in the mu heavy chain: partial deletion, polypeptide addition, and abnormal glycosylation have been observed. These mutant cell lines provide a means of identifying the structural basis of IgM function and of studying the biochemistry of IgM synthesis and processing.     

The disulfide bonding pattern in ficolin multimers

Ficolin is a plasma lectin, consisting of a short N-terminal multimerization domain, a middle collagen domain, and a C-terminal fibrinogen-like domain. The collagen domains assemble the subunits into trimers, and the N-terminal domain assembles four trimers into 12-mers. Two cysteine residues in the N-terminal domain are thought to mediate multimerization by disulfide bonding. We have generated three mutants of ficolin alpha in which the N-terminal cysteines were substituted by serines (Cys4, Cys24, and Cys4/Cys24). The N-terminal cysteine mutants were produced in a mammalian cell expression system, purified by affinity chromatography, and analyzed under nondenaturing conditions to resolve the multimer structure of the native protein and under denaturing conditions to resolve the disulfide-linked structure. Glycerol gradient sedimentation and electron microscopy in nondenaturing conditions showed that plasma and recombinant wild-type protein formed 12-mers. The Cys4 mutant also formed 12-mers, but Cys24 and Cys4/Cys24 mutants formed only trimers. This means that protein interfaces containing Cys4 are stable as noncovalent protein-protein interactions and do not require disulfides, whereas those containing Cys24-Cys24 require the disulfides for stability. Proteins were also analyzed by nonreducing SDS-PAGE to show the covalent structure under denaturing conditions. Wild-type ficolin was covalently linked into 12-mers, whereas elimination of either Cys4 or Cys24 gave dimers and monomers. We present a model in which symmetric Cys24-Cys24 disulfide bonds between trimers are the basis for multimerization. The model may also be relevant to collectin multimers.