A peptide extension dictates IgM assembly
- aCenter for Integrated Protein Science Munich at the Department Chemie, Technische Universität München, 85748 Garching, Germany;
- bDivision of Genetics and Cell Biology, Università Vita-Salute IRCCS Ospedale San Raffaele, 20132 Milano, Italy;
- cFakultät Chemie, Institut für Biologische Chemie, Universität Wien, 1090 Wien, Austria
See allHide authors and affiliations
Edited by Linda L. Randall, University of Missouri-Columbia, Columbia, MO, and approved September 1, 2017 (received for review February 1, 2017)

Significance
How protein assemblies with complex topologies are formed is an important question in structural biology. An intriguing example is IgM, a complex of over 1,200 kDa consisting of six antibody subunits (or five in the presence of the J-chain protein). These are arranged in a ring-like structure connected by disulfide bonds. Here, we show that in vitro and in cell culture, a short peptide extension of the IgM heavy chain is sufficient to steer the formation of the hexameric complex. The formation of a disulfide bond triggers conformational changes in the peptide extensions, which involve specific hydrophobic residues. Our study reveals the redox-controlled assembly of a large protein complex via structural rearrangements in a peptide as a design principle.
Abstract
Professional secretory cells can produce large amounts of high-quality complex molecules, including IgM antibodies. Owing to their multivalency, polymeric IgM antibodies provide an efficient first-line of defense against pathogens. To decipher the mechanisms of IgM assembly, we investigated its biosynthesis in living cells and faithfully reconstituted the underlying processes in vitro. We find that a conserved peptide extension at the C-terminal end of the IgM heavy (Ig-μ) chains, termed the tailpiece, is necessary and sufficient to establish the correct geometry. Alanine scanning revealed that hydrophobic amino acids in the first half of the tailpiece contain essential information for generating the correct topology. Assembly is triggered by the formation of a disulfide bond linking two tailpieces. This induces conformational changes in the tailpiece and the adjacent domain, which drive further polymerization. Thus, the biogenesis of large and topologically challenging IgM complexes is dictated by a local conformational switch in a peptide extension.
Secretory IgMs are polymeric antibodies that provide a first-line defense in vertebrates against invading microorganisms and other pathogens (1). Like in other Ig classes, two μ heavy (H) and two light (L) chains assemble covalently into μ2-L2 subunits (often also referred to as “monomers” in the Ig context). The assembly pathways vary in different isotypes. In IgM, μ-L assembly precedes dimerization, while in IgG1, H dimerization precedes assembly with L (2, 3). In plasma cells, IgM monomers are further arranged into (μ2-L2)6 hexamers in the absence of the J chain, or into (μ2-L2)5 J pentamers in its presence (4, 5). The μ chains can form oligomers that are able to incorporate the J chain in the absence of L chains. However, μ oligomers are binding immunoglobulin protein (BiP)-associated, and are therefore not secreted (6). Abundant μ2 intermediates accumulate in plasma cells, with their further polymerization being limited by efficient endoplasmic reticulum (ER)-associated degradation (7).
The polymeric structure increases the avidity for antigen and complement factors such as C1q, compensating for the lower affinity of most IgM (8, 9). The IgM H (Ig-μ) chains contain four constant domains (Cμ1–Cμ4) with no hinge region. The Cμ4 domain harbors a C-terminal extension, the so-called μ tailpiece (µtp), which is essential for polymerization (10⇓⇓–13). A related peptide is found in IgA, an isotype that also binds J chains and forms oligomers (12, 14).
In polymers, μ chains are connected by several interchain disulfide bridges. The one formed between the Cμ2 domains (C337) stabilizes μ2-L2, while C414 in Cμ3 and C575 in the μtp are involved in intersubunit disulfide bonds (10, 15⇓–17). In addition to C337-dependent covalent bonds, μ2-L2 subunits can be stabilized by noncovalent interactions (17, 18).
Among the cysteines, C575 in the μtp is of special importance, since its replacement with either alanine or serine has drastic effects on polymerization (12, 17). Furthermore, C575 mediates IgM quality control and degradation (19⇓⇓–22). In these thiol-mediated processes (12, 14, 23, 24), C575 of unassembled IgM covalently interacts with C29 of ERp44, a protein disulfide isomerase (PDI) family member, which retrieves incomplete IgM oligomers from the Golgi back to the ER (20, 25). C575 is also the site for the covalent attachment of J chains to IgM and IgA polymers (26⇓–28).
We have previously solved the structures of all individual IgM constant domains at atomic resolution and presented a model for the IgM oligomer (13). We showed that the Cμ4 domain and the μtp are necessary and sufficient for the specific polymerization into hexamers of covalently linked dimers (i.e., dodecamers). However, the molecular mechanism of how the μtp promotes oligomerization was still unclear. In this study, we determined the principles underlying polymer formation and, by replacing every amino acid in the μtp, we identified the residues essential for polymer assembly in vitro and in cell culture.
Results
The μtp Dictates Oligomerization of the Entire IgM.
The regulated formation of IgM hexamers is still a mysterious process. Previous work has shown that the Cμ4 domain extended by the 18-aa C-terminal μtp is sufficient to drive hexamer formation (13). To define the role of the μtp amino acids in the polymerization of the entire IgM, we replaced every single residue in the μtp by alanine and analyzed the effects of these and other point mutations in cultured cells and in vitro.
For the cell culture analyses, the point mutations were inserted into full-length Ig-μ and coexpressed in HEK-293T cells with Ig-λ L chains to form 4-hydroxy-3-nitrophenyl acetyl–binding IgM assemblies (12). The molecular composition of intracellular and secreted subunits was analyzed by Western blotting with anti-μ and anti-λ antibodies (Fig. 1A and Fig. S1B). The four mutants with a markedly reduced hydrophobicity (Y562A, V564A, L566A, and I567A), were secreted exclusively as μ2-L2 “monomers.” In addition to the more abundant monomers, M568A and C575A transfectants secreted some covalent complexes of higher molecular weight (MW). Other mutants with increased hydrophobicity (N563A, S565A, and D570A), the former two of which destroy the glycosylation site characterized by the amino acid sequence NVS, secreted mainly hexamers and higher MW species.
Effects of μtp alanine replacements on IgM polymerization in cultured cells. (A) Cell SNs were collected after 4 h of incubation in MEM and resolved by nonreducing gels, and blots were decorated with Ig-μ antibodies. Owing to unbalanced μ-L stoichiometries, the relative amounts of μ, μ-L, μ2-L, and μ2-L2 assemblies varied in different transfectants. (B) Different gel mobility of mutants secreted as μ2-L2 is evident under reducing (a) and nonreducing (b) conditions.
Characterization of IgM alanine mutants. (A) Sucrose gradient fractionation assays confirm that mutants with reduced hydrophobicity are secreted as μ2-L2 subunits. Cell SNs were processed as described in Materials and Methods. Eight of the total 12 fractions (FR.) collected were resolved by nonreducing polyacrylamide gradient gels. Polymeric wt IgM accumulates in fractions 7–8, while Y562, V564, L566, and I567 mutants are present mostly in fractions 2–4. These results exclude the formation of polymers. Note that the M568A mutant yields both polymers and monomers. (B) Cell lysates were collected after 4 h of incubation in MEM and resolved by nonreducing gels, and blots were decorated with Ig-μ (Top) or Ig-λ (Bottom) antibodies. Different assembly intermediates (μ, μ-L, μ2-L, and μ2-L2) are indicated on the right-hand margin. The top two bands are hexamers and pentamers. IgM Std, IgM standard.
The μ2-L2 complexes secreted by Y562A, V564A, L566A, I567A and M568A could have been part of bigger noncovalent oligomers, which were dissociated during SDS/PAGE. To test this possibility, the corresponding supernatants (SNs) were fractionated by sucrose gradient density centrifugation. No signal was detected in fractions 6–8, where covalent wild-type (wt) IgM polymers migrate (Fig. S1A). Thus, we can exclude the secretion of noncovalent polymers for these mutants, as well as for C575A (16). Further evidence for their inability to form oligomers was obtained by SDS/PAGE analyses of the intracellular fractions. Few, if any, intermediates of higher MW than (μ2-L2)2 dimers could be observed for the hydrophobic mutants (Fig. 1A).
With the exception of Y562A, V564A, L566A, I567A, M568A, and C575A, the other mutants were secreted mainly as pentamers and hexamers in different proportions. In cell lysates, IgM formed a smear in the upper part of the gels. These high-MW species were present also in M568A and C575A (Fig. 1A and Fig. S1B). These findings highlight the stringency of the IgM quality control mechanisms. These ensure that, on the one hand, only oligomers of correct size are secreted and, on the other hand, that all intermediates, including μ2-L2 monomers, are prevented from secretion (12, 25). Therefore, the mutants secreted as μ2-L2 monomers must be somehow able to escape the thiol-dependent control machinery.
Owing to their differential accessibility to the modifying enzymes in the Golgi complex, the processing of the N563 glycans depends on the oligomeric state of IgM transiting through the secretory pathway (11, 29). Thus, C575A Ig-μ chains show slower electrophoretic mobility under reducing conditions due to extensive processing of the N563 glycans by Golgi enzymes (29). When analyzed by reducing gels, the L566 mutant secreted exclusively as μ2-L2 displayed slow mobility (Fig. 1B). The M568A mutant, which was secreted, in part, also as pentamers and hexamers, migrated instead as a doublet under reducing conditions. Interestingly, when the same samples were run under nonreducing conditions, L566A and M568A showed a faster electrophoretic mobility than C575A (Fig. 1B), suggesting that mutants in which a hydrophobic residue was replaced adopted a more compact conformation. One possibility is that formation of C575-dependent disulfides within the same subunit reduces their hydrodynamic volume. The presence of an additional intrasubunit bond may also explain the paucity of secreted μ-L “hemimers” in the SNs of L566A and M568A transfectants.
Alanine replacement of either N563 or S565, two mutations that prevent attachment of the highly conserved N-glycan, caused secretion of hexameric and higher MW complexes. The N563 glycans serve important roles in determining the extent and velocity of polymerization (20). To further investigate the influences of glycosylation on polymer size, the mutant N563Q was generated, as glutamine closely resembles the wt asparagine, but N-glycosylation is no longer possible (Fig. 1A). N563Q formed fewer high-MW species than N563A or S565A. The stronger aggregation tendency of mutants with an A than Q at position 563 may reflect their higher overall hydrophobicity. To mimic the environment normally encountered in plasma cells, we coexpressed J chains with WT, L566A, M568A, or C575A μ chains. No polymers or intermediates containing J chains were secreted by the mutants (Fig. S2). Thus, cysteine 575 is not reactive with J chains in the mutants with increased hydrophobicity.
Mutants with decreased hydrophobicity do not bind J chains. HEK-293T cells stably expressing l chains were cotransfected with vectors encoding murine J chain and the indicated μ-chain mutants. Aliquots of the SNs were resolved under nonreducing conditions, and the blot was sequentially decorated with anti-J and anti-μ antibodies (6).
Cμ4tp Forms Cμ4tp2 Hexamers via a Trimeric Intermediate in Vitro.
To achieve more detailed insight into IgM assembly, we performed in vitro experiments with the minimum oligomerization module, the Cμ4 domain with the attached μtp extension (i.e., Cμ4tp) (13). In vitro, the Cμ4tp domains associated into hexamers of covalent dimers [(Cμ4tp2)6] without any additional factors. To identify potential intermediates, we followed the association of the Cμ4tp domains by analyzing the process at defined time points by size exclusion (SE)-HPLC (Fig. 2A).
Kinetics of Cμ4tp oligomerization in vitro. Cμ4tp was refolded and purified as described in Materials and Methods. SEC fractions containing the Cμ4tp monomer were pooled, and the concentration was adjusted to 1 mg/mL. Aliquots were stored at 4 °C and analyzed at the indicated time points. (A, a) Representative SE-HPLC profiles on a Superdex 200 10/300 GL in PBS at 20 °C. For each time point, 100 μg of sample was injected. A BioRad gel filtration standard (gray) was used as a size reference, where peak 2, which is the closest to the monomer, refers to myoglobin (horse) with a molecular mass of 17 kDa. (A, b) Monomer (red), dimer (orange), and dodecamer (green) formation is shown over time. Error bars indicate the SD of three independent SE-HPLC measurements. (B) SDS/PAGE analysis under nonreducing conditions of the kinetic samples blocked with AMS at indicated time points. Relative optical (Opt.) density of the monomer (red bar) and covalent dimer (orange bar) bands over time is shown. Error bars represent the SD of the optical density of three individual nonreducing SDS/PAGE experiments.
The chromatograms indicated the presence of four main species: the Cμ4tp monomer, Cμ4tp dimers (Cμ4tp2), an intermediate species corresponding to trimers of dimers (i.e., Cμ4tp hexamers [(Cμ4tp2)3]), and hexamers of dimers (i.e., Cμ4tp dodecamers [(Cμ4tp2)6]). These species were present over the period of the analysis, but their relative abundance changed. Whereas monomers progressively decreased over the time course of the experiment, dodecamers increased. Cμ4tp2 and (Cμ4tp2)3 reached a plateau early on (∼37% and ∼2%, respectively), and remained almost constant, suggesting that those species are intermediates in the polymerization process. Our data show that a certain Cμ4tp2 threshold concentration is needed to allow hexamer formation. Nonreducing SDS/PAGE analyses revealed the presence of only two species: monomers, whose intensity gradually decreased during the experiment, and covalent dimers, whose intensity increased (Fig. 2B). Together, these results clearly indicate that the end product [i.e., (Cμ4tp2)6, the (Cμ4tp2)3 intermediate] are noncovalent assemblies of the first intermediate to appear, the covalent dimer (i.e., Cμ4tp2).
Formation of C575 Disulfide Bonds Is a Prerequisite for Oligomerization in Vitro.
To investigate whether and how redox conditions affected oligomerization, reduced glutathione (GSH) and oxidized glutathione (GSSG) were added to the isolated Cμ4tp monomer at the beginning of the kinetics. Upon addition of the redox system, not only the formation of the covalent dimers but also that of intermediates and (Cμ4tp2)6 was arrested (Fig. S3). These data suggested that the disulfide bond between C575 of two Cμ4tp monomers cannot stably form in the presence of GSH and that this bond plays a decisive role in the oligomerization process.
Effect of the redox system during oligomerization. Cμ4tp was refolded under the standard GSSG/GSH ratio. The monomer was isolated as outlined above, and the protein concentration was adjusted to 1 mg/mL. The sample was then dialyzed “ON” against PBS containing three different ratios of the redox system. Aliquots were stored at 4 °C and analyzed at the indicated time points by SE-HPLC on a Superdex 200 10/300 GL in PBS at 20 °C. For each time point, 100 μg of sample was injected. Black bars indicate no GSSG/GSH, yellow bars indicate GSSG/GSH = 1 mM/0.5 Mm, gray bars indicate GSSG/GSH = 1 mM/1 Mm, and green bars indicate GSSG/GSH = 0.5 mM/1 mM.
To confirm that covalent dimerization of Cμ4tp is essential for further oligomerization, we monitored the kinetics of assembly under conditions in which formation of C575 bonds was not possible (Fig. 3 A–C). To this end, we analyzed the association properties of Cμ4tp monomers in which C575 was irreversibly blocked by 4-acetamido-4′-maleimidylstilbene-2,2′-disulfonic acid (AMS) by SE-HPLC (Fig. 3A). No dimers became detectable even at a high concentration (15 mg/mL), indicating that the oligomerization process was not initiated when the formation of covalent Cμ4tp2 species was prevented.
Essential role of C575 disulfides in Cμ4tp oligomerization in vitro. (A and B) Effects of AMS on Cμ4tp oligomerization. (A) AMS (10 mM final) was added to freshly isolated monomeric Cμ4tp, and the sample was concentrated to 15 mg/mL. (B) AMS was added after 0 (black), 1 (red), 2 (blue), 3 (pink), 6 (green), or 9 (purple) d of incubation at room temperature, and the samples were analyzed at the indicated time points. (C) Defective oligomerization of C575 Cμ4tp mutants: C575S and C575A monomer was isolated and concentrated to 1 mg/mL (red) or 15 mg/mL (black) directly in PBS or after dialysis against PBS with 1 M NaCl. (D) Representative SE-HPLC experiments show the disruption of the C575 disulfide bonds by DTT, which causes (Cμ4tp2)6 dissociation. Dodecamers were obtained by processing the concentrated Cμ4tp over the preparative SE column. Fractions containing (Cμ4tp2)6 were pooled, the protein concentration was adjusted to 1 mg/mL, and DTT (1 mM final) was added. Samples were analyzed at the indicated time points. Green bars indicate (Cμ4tp2)6, orange bars indicate Cμ4tp2, and red bars indicate Cμ4tp. All SE-HPLC runs were performed injecting 100 μg of sample on a Superdex 200 10/300 GL column at 20 °C. Error bars represent the SD of three independent SE-HPLC measurements.
Second, we added an excess of AMS at different time points during oligomerization, thus blocking the remaining free C575 (Fig. 3B). Clearly, upon irreversible modification of C575, no further conversion of Cμ4tp monomers into (Cµ4tp2)6 was observed over time. This implies that both dimerization and further oligomerization were blocked. The amount of Cμ4tp monomers and (Cμ4tp2)6 present before the addition of AMS remained unchanged.
Altogether, these data are strong evidence for the formation of covalent dimers as the first and essential step for further noncovalent oligomerization into (Cμ4tp2)6.
Third, we tested whether noncovalent interactions could drive oligomerization in the absence of C575, comparing two mutants in which the cysteine was replaced with serine (isosteric and polar) or with alanine [isovolumetric and hydrophobic (30)]. At all protein concentrations investigated, for both the C575A and C575S mutants, the main species (96–99%) was the monomer (Fig. 3C). Also, the addition of 1 M NaCl (Fig. 3C), which enhances hydrophobic interactions, did not result in a detectable increase in higher oligomeric species. Taken together, these results indicate that stable noncovalent interactions, which lead to oligomerization, can be established only after formation of covalent linkages between C575 residues.
Finally, to assess how stable the noncovalent interactions are within (Cμ4tp2)6 after disruption of the C575 S-S bond, we treated the dodecamers with a reducing agent (1 mM DTT) and analyzed the oligomeric state by SE-HPLC at different time points. As shown in Fig. 3D, after addition of DTT (Cμ4tp2)6 decreased progressively, while monomers increased to become the prevalent species. Thus, by disrupting the C575 S-S bonds, noncovalent interactions are not able to persist.
Hydrophobic μtp Residues Are Essential for Oligomerization of Covalent Cμ4tp2 in Vitro.
Unlike most other Ig CH domains, Cμ4 is prevalently monomeric in solution (13, 31). Since Cμ4tp fragments assemble into (Cμ4tp2)6 complexes (13) (Fig. 3), the 18-aa-long μtp is the element that confers the ability to oligomerize, likely allowing formation of alternate covalent and noncovalent linkages with adjacent Cμ4tp. So far, we assessed the effects of the μtp extension on IgM polymerization in cultured cells. Furthermore, we established in vitro that the penultimate cysteine is important for assembly. To verify our cell culture results and to gain a deeper understanding of the structural bases leading to oligomerization, each μtp amino acid in the Cμ4tp context was replaced by alanine, and the effects of these mutations on oligomerization were then analyzed in vitro.
All alanine mutants, except C575A, were able to form covalent dimers. Based on their ability to oligomerize, the mutants can be grouped in three categories: (i) mutants with no ability to oligomerize at all, (ii) mutants capable of oligomerization, and (iii) mutants with a tendency to aggregate (Fig. 4A, Fig. S4A, and Table 1). Interestingly, mutants that formed covalent dimers but completely lost the ability to assemble (Y562A, V564A, L566A, and I567A) were those where a hydrophobic residue was replaced by alanine (Fig. 4B). This perfectly matches our cell culture studies and supports the fact that hydrophobic μtp residues are essential for oligomerization. To assess whether loss of oligomerization could be overcome in an environment that enhances hydrophobic interactions, mutants were analyzed in the presence of 1 M NaCl (Fig. S4B). Under these conditions, only I567A showed a detectable oligomerization tendency at higher protein concentrations. Thus, the effects of mutating single hydrophobic residues could not be compensated by favoring hydrophobic interactions.
Oligomerization ability of Cμ4tp alanine mutants in vitro. (A) Schematic representation of their oligomerization ability in vitro. Mutants were refolded, and monomers were isolated as previously described. To accelerate oligomerization, monomers were concentrated to 15 mg/mL and then run over a preparative SE column. Fractions containing the oligomer were pooled, and the concentration was adjusted to 0.5 mg/mL for analytical ultracentrifugation (aUC) and to 1 mg/mL for SEC-MALS analysis. When no oligomers formed, fractions containing monomer and dimers were pooled and analyzed at the same concentrations. For mutants able to oligomerize the height of the dodecamer, the peak was normalized with respect to that of the wt dodecamer peak. Red indicates Cμ4tp, orange indicates Cμ4tp2, yellow indicates (Cμ4tp2)5, and green indicates (Cμ4tp2)6. (B) SEC-MALS (a) and aUC (b) profiles of nonoligomerizing mutants in PBS. c(s), concentration of species as a function of sedimentation; Sed. coefficient, sedimentation coefficient. (C) Schematic representation of the region prone to β-aggregation predicted by TANGO and by AGGRESCAN of the wt μtp peptide and 18 alanine μtp mutants.
Nonoligomerizing, oligomerizing, and aggregating Cμ4tp mutants with the molecular mass and S value of the biggest species formed
Oligomerization ability of Cμ4tp alanine mutants in vitro. (A, a and b) SEC-MALS (Left) and the aUC profiles of oligomerizing mutants in PBS (Right). (A, c) Chromatograms of the preparative SEC of aggregating mutants. aUC, analytical ultracentrifugation; Sed. coeffcient, sedimentation coefficient. (B) Oligomerization ability of nonoligomerizing mutants at 15 mg/mL. The samples were analyzed on SE-HPLC injecting 100 μg of protein. (C) TANGO and AGGRESCAN scores for aggregation propensity of μtp mutants.
A detailed analysis of the mutants revealed interesting features. P559A, T560A, and L561A exhibited a severely compromised ability to form polymers in vitro (Fig. 4A and Fig. S4A). The hydrophobic mutant M568A formed mainly decamers (Cμ4tp2)5 of 168 kDa (Table 2). It should be recalled that the covalent dimerization of Ig-μ observed in HEK-297T cells depends primarily on intrasubunit disulfide bonds between C337 in the Cμ2 domain.
Synthetic μtp peptides
Two mutants, N563A and D570A, displayed a strong tendency to aggregate, as also observed in cultured cells. In N563A, the attachment site for N-glycosylation is mutated. The effects of glycosylation at position 563 on polymerization were suggested above. D570A lacks the only negative charge present on the μtp, which might act as a gatekeeper that prevents aggregation (32⇓–34). Interestingly, with the exception of C575A, all mutants showing defective oligomerization carry a mutation in the N-terminal half of the μtp, spanning residues 559–570. To gain further information on their biophysical properties, we interrogated programs designed to predict the β-aggregation propensity of polypeptide sequences (35, 36). Analysis of the wt μtp and its Ala mutants identified a region located in the N-terminal half of the sequence, encompassing residues 559–569 (TANGO) and 559–570 (AGGRESCAN), which influences the β-aggregation tendency of the μtp (Fig. 4C). Both programs predicted a strikingly higher aggregation propensity for the N563A mutant and a significantly lower aggregation propensity for L561A, Y562A, V564A, L566A, and I567A (Fig. S4C). Remarkably, with the exception of C575A, the behavior of all IgM and Cμ4tp mutants was in line with TANGO and AGGRESCAN predictions. Thus, residues in the μtp dictate the extent of IgM oligomerization.
Replacement of Hydrophobic Residues Alters the μtp Secondary Structure.
To assess the effects of μtp mutations on the isolated μtp, we synthesized 18-aa-long peptides encompassing residues 559–576 and carrying the respective point mutations. The μtp mutants were tested for oligomerization, either in the absence or presence of a redox system (GSSG/GSH) and analyzed by SE-HPLC (Fig. 5A). All mutants except C575A and C575S formed covalent dimers upon oxidation. However, not even the wt μtp was able to assemble further into oligomers, suggesting that interactions between Cμ4 and μtp of the same chain or adjacent chains are important for polymerization. When μtp monomers or dimers were mixed with Cμ4 and incubated, no oligomeric species were detected, indicating that in cis, μtp-Cµ4 continuity is required for oligomerization.
Characterization of wt and mutant μtp peptides. (A) Oligomerization ability of the wt μtp peptide. SEC profiles of μtp peptides resuspended before (a) or after (b) oxidation on a Superdex 75 10/300 GL column on day 0 (red) and day 3 (black) after resuspension/oxidation are shown. In a, the peptide was resuspended in Tris/EDTA buffer (pH 7.5) to a final concentration of 0.4 mg/mL (207 μM) and stored at room temperature (RT). In b, a redox system was added (1 mM GSSG, 0.5 mM GSH final) and the sample was incubated for 2 h at RT under stirring. Glutathione was then removed by centrifugation/filtration and subsequent dilution with Tris/EDTA buffer. Samples were stored at RT. (B and C) Secondary structure characterization of resuspended/oxidized wt μtp within the far-UV range. (B, a) Overlay of the CD spectra of the resuspended (green) and oxidized (red) wt peptide. Spectra were acquired at 20 °C. (B, b) PSIPRED prediction of the wt μtp secondary structure. (C) Overlay of the CD spectra of four μtp mutants after oxidation. Red indicates wt, black indicates Y562A, blue indicates V564A, pink indicates L566A, and green indicates I567A.
The secondary structure of the peptides was assessed by far-UV CD spectroscopy. The wt peptide showed a CD spectrum consistent with a β-strand motif (Fig. 5B, a). Upon oxidation, the CD spectrum changed, yielding a more pronounced minimum of ellipticity at 215 nm. In agreement with TANGO and AGGRESCAN, the PSIPRED prediction for secondary structure (bioinf.cs.ucl.ac.uk/psipred/) suggests the presence of a β-strand motif on the N-terminal half of the μtp from 560 to 569 (Fig. 5B, b). The mutants in which hydrophobic residues were substituted exhibited different characteristics (Fig. 5C). These hydrophobic residues are located in the stretch from 560 to 569, which contributes to the β-strand according to the PSIPRED prediction. Interestingly, all IgM and Cμ4tp mutants that showed a significantly altered oligomerization behavior were those carrying mutations within this stretch. Taken together, these findings indicate that the hydrophobic residues of the μtp are key for the formation of a structure required for oligomerization in the context of the Cμ4 domain. An altered conformation might thus explain the fate of aggregating and nonoligomerizing mutants.
C575 Reactivity and Y576 Hydrophobicity Control IgM Polymerization.
Next, we assessed the impact of the C-terminal doublet, CY. When Y576 was replaced with alanine, the oligomerization ability of both IgM in HEK-297T cells and Cμ4tp in vitro was significantly affected (Figs. 1A, 4A, and 6). To determine whether higher hydrophobicity at this position would instead favor oligomerization, Y576 was replaced with residues of increasing hydrophobicity: glycine, tryptophan, or phenylalanine. In both Cμ4tp and IgM, an increased hydrophobicity of the C-terminal residue promoted oligomerization, mainly (μ2-L2)5 pentamers being formed in cultured cells (Fig. 6A, a and B, b).
Role of the C-terminal part of the μtp in cell culture and in vitro polymerization. (A, a and b) Secreted IgM was collected from the indicated mutants and analyzed as described in the legend for Fig. 1. (B) In vitro oligomerization of Cμ4tp mutants. (Left) SEC-MALS profiles of isolated oligomers at 1 mg/mL in PBS. (Right) Analytical ultracentrifugation (aUC) profiles at 0.5 mg/mL in PBS. (a) Red represents wt, and black represents ΔY576. (b) Red represents wt, black represents Y576G, cyan represent Y576A, purple represents Y576W, and orange represents Y576F. (c) Red represents wt, blue represents Y18K, and yellow represents Y18D. (d) Red represents wt, and black represents C575YY576C. Sed. coefficient, sedimentation coefficient.
To test whether and how the presence of a charged residue at this position affects polymerization, we replaced Y576 with either lysine or glutamate (Fig. 6A, b and B, c). The presence of a positive charge (Y576K) supported the formation of hexameric Cμ4tp2 and μ2-L2. The Y576D mutant carrying a negative charge close to C575 had opposite effects. In the Cμ4tp context, no oligomerization occurred and the efficiency to form covalent dimers was reduced, most likely due to the repulsion of the negative charges in the adjacent positions 575 and 576. These results could also reflect effects on the pKa of C575, and hence its reactivity. In the IgM context, a negative charge in the last position favored secretion of μ2-L2 subunits, as did also upstream C575 (14).
Finally, to investigate the relevance of the position of the cysteine within the μtp, and therefore the position of disulfide bonds within the polymers, we generated mutants in which the cysteine was placed at the very C-terminal position or at increased distances from the C terminus. When the CY sequence was inverted (C575YY576C), both the Cμ4tp and the IgM mutants formed polymers. In particular, the YC mutant formed more Cμ4tp2 hexamers than the wt in vitro, while in the IgM context, hexamers were favored over pentamers (Fig. 6A, b and B, d). In contrast, the three Cμ4tp double-point mutants, where C575 was replaced by serine and a cysteine was inserted at position 559, 565, or 569, displayed a stronger propensity to aggregate. Thus, shortening the distance between the μtp cysteine and the Cμ4 domain is deleterious to oligomerization.
Taken together, these experiments reveal further constraints for the composition of the μtp and the positioning of specific residues.
Discussion
Our data reveal that the C-terminal μtp contains crucial information for IgM assembly, which matches its evolutionary conservation. In a previous study, we had demonstrated that extending the Cμ4 with a μtp is sufficient to drive the formation of hexamers of dimers in vitro without any additional factors (13). In this work, we analyzed the underlying mechanism by determining the role of individual residues within the μtp. An important conclusion is that disulfide linkage between the penultimate cysteines (C575) of two Cμ4tp monomers is a prerequisite for assembly. Only after this covalent connection occurs can noncovalent interdimer associations start and continue until hexamers of covalent (Cμ4tp)2 are formed (Fig. 7).
Models for IgM polymerization. (A) Model for Cμ4tp oligomerization in vitro. (a and b) μtp (dark blue tail) conformational changes in Cμ4tp induced by the formation of a C575 disulfide bond (yellow circles). In a, two different possible tailpiece conformations are depicted. Cμ4tp dimerization is coupled to oligomerization (b and c), leading to a hexamer (d). Noncovalent interactions might occur either between two adjacent μtps or between the μtp and the Cμ4 domain in trans. (B) Assembly of IgM in living cells. Cysteines involved in interdomain disulfide bridges are indicated by yellow circles. (a) Fc region of two μ-L “hemimers” is shown with one of the two μtp arrangements shown in A. Covalent and noncovalent interactions link C337 and surrounding regions in the Cμ2 domains (depicted here in green), limiting the mobility of the C-terminal portions, and hence favoring collisions between pairs of Cμ4 and μtps, to produce a μ2-L2 as depicted in b. Antigen-binding (Fab) fragments attached to the Cμ2 domains were deleted for clarity. (b) Snippet of the IgM Fc dodecamer. Formation of the C575 disulfide results in conformational modifications the Cμ4tp domain, allowing the establishment of noncovalent interactions (red boxes) either between two adjacent μtps or between the Cμ4 and a μtp in trans. The Cμ3 domains in adjacent subunits must become close enough for C414 to form a disulfide bond, a feature favored in hexamers. It is noteworthy that C414 does not form intersubunit disulfides in membrane IgM.
Strikingly, the results obtained for the assembly of tailpiece mutants were almost identical in vitro and in cell culture. The effects observed for Cμ4tp in vitro faithfully phenocopied those for IgM in living cells, demonstrating directly that all of the information required for IgM assembly is contained in this module. However, the slow kinetics of oligomerization in vitro support the notion that additional intrinsic factors like other domains or N-glycosylation and/or cellular factors such as ERp44 and ERGIG53 (25) further assist IgM biogenesis in cultured cells.
While impeding the formation of the C575-disulfide bond in vitro prevents oligomerization, none of the mutations that abolish further noncovalent assembly prevent the covalent linkage. Importantly, this indicates that covalent association of two Cμ4tp monomers via the tailpiece must precede the establishment of noncovalent interactions leading to dimerization and oligomerization. Since (Cμ4tp2)6 formation is topologically quite challenging and the amount of dimers stays constant in vitro, this suggests that a certain Cμ4tp2 threshold concentration is needed to allow formation of the dodecamer. The mutations severely affecting oligomer formation are replacements of L561A, V564A, L566A, M568A, and Y576A, with Y562A and I567A completely hindering assembly. A likely explanation is that these hydrophobic residues are involved in different interactions in the absence of the C575 disulfide bond. Thus, formation of the disulfide bridge seems to induce rearrangements within the Cμ4tp, which impose steric constraints that, in turn, allow hexamer formation. However, oligomerization is possible only when the μtp is anchored to the Cμ4 domain: Isolated μtp exclusively formed covalent dimers, and no further association was observed when Cμ4 and μtp were incubated in trans. Our data do not allow us to discriminate whether the noncovalent interactions leading to oligomerization occur between two μtps or between a μtp and Cμ4 domain (Fig. 7). Interestingly, the peptides generated upon mutation of the highly hydrophobic residues seem to be less structured than the wt μtp. In cultured cells, the mutants show increased electrophoretic mobility under nonreducing conditions, pointing to a more compact conformation, possibly due to intrasubunit C575 disulfide bonding. This bond might be formed by wt μ chains as well, but recognized as nonnative and isomerized into intersubunit linkages by ERp44 or other oxidoreductases of the early secretory pathway.
The CD spectrum of the wt μtp peptide presents features of a β-strand motif that, according to the PSIPRED algorithm, is located in the N-terminal half of the μtp, encompassing residues 559–568. Also TANGO and AGGRESCAN predicted an increased aggregation propensity for these mutants. Mutations in this region severely compromised oligomerization (i.e., no oligomerization at all or aggregation). Thus, our in vitro experiments reveal that the reduced μtp is structured and that this structure is rearranged upon oxidation as the starting point for interactions involving the μtp and the Cμ4 domains that are needed for assembly. Importantly, moreover, the in vitro behavior of the Cμ4tp mutants was identical to that of the corresponding full-length IgM mutants in HEK-297T cells. This proves that the hydrophobic μtp amino acids are essential for IgM oligomerization and that intracellular cofactors cannot correct this.
It is of special interest that mutants in the hydrophobic core (Y562A, V564A, L566A, I567A, and M568A) are secreted from cells as μ2-L2 subunits. This is surprising because thiol-mediated retrieval normally prevents the secretion of incomplete or incorrectly assembled polymers (16). ERp44 (25, 37) and the C-terminal cysteine C575 (12, 22) are key players in this process. Thus, the essential element required for oligomer formation resides in the region mutated. In these mutants, ERp44 could be unable to recognize C575 due to a collapse of the two μtps that reduces their accessibility and possibly by the formation of intrasubunit C575 disulfide bonds. In cells, the μtps within the same subunit are in close proximity due to the interactions and the covalent bond linking the Cμ2 domains (13, 17, 18). Mechanisms must hence operate that prevent formation of intrasubunit C575 bonds, or isomerize them into native intersubunit interactions.
Our results also shed new light on the effects of glycosylation. The number of subunits in IgM polymers is influenced by the presence of glycan moieties linked to N563. N563A (this work) and S565A (ref. 38 and this work), which lack them, are secreted as hexameric and higher MW species. The presence of glycans in the tightly packed IgM core (13, 39) could limit the number of subunits that can be incorporated into an oligomer or serve as a docking device for ERGIC53, a hexameric lectin shown to promote polymerization in nonlymphoid cells (25). The N563Q mutant tended to form fewer aggregates. Furthermore, of the 18 Cμ4tp mutants analyzed in vitro, all being nonglycosylated, only two formed aggregates (N563A and D570A). This indicates that the increased aggregation propensity of N563A, S565A, and D570A (as predicted by TANGO and AGGRESCAN) is responsible for the formation of high-MW species. N563 glycosylation might be relevant for J-chain incorporation and the formation of pentameric IgM.
Based on our data and the literature, we propose a model for IgM assembly in which the geometry of IgM assembly is determined by the Cμ4 domains together with the tailpiece (Fig. 7). As a committed step, monomeric Cμ4 domains (in isolation or in the context of the H chain) are covalently linked via C575 disulfide bridges. This induces structural changes involving hydrophobic residues in the μtp as essential factors for IgM assembly. These rearrangements trigger events leading to the noncovalent association of the Cμ4tp domains and the oligomerization into hexamers. Translated to full-length IgM, this implies that formation of C575 disulfides is required for the formation of intersubunit covalent and noncovalent interactions. In fact, without C575, μ2-L2 subunits are the main secreted species, with few covalent polymers formed via C414. Unexpectedly, these species are absent in the hydrophobic μtp mutants, suggesting an important role for the N-terminal half of the μtp in forming defined polymers. One possibility is that intrasubunit C575 disulfides are formed first as the two μ chains in IgM subunits are linked by C337 disulfides. The intrasubunit C575 disulfides could be isomerized with the help of ERp44 and other cellular factors. The rearranged hydrophobic tailpiece residues would then mediate formation of planar polymers in which isomerization can take place. Thus, this small extension, together with the Cμ4 domain, orchestrates the sophisticated assembly of large IgM complexes.
In the context of engineering antibodies with enhanced effector function (40, 41), our work provides a rational basis for the development of peptide appendixes able to confer specific oligomerization properties. This would allow turning antibodies into polyvalent or multispecific entities that could be employed as therapeutic agents with improved effector functions.
Materials and Experimental Procedures
Materials.
Unless otherwise specified, all experiments in vitro were carried out in PBS [8.09 mM Na2HPO4, 1.76 mM KH2PO4, 137 mM NaCl, and 2.65 mM KCl (pH 7.4)] at 20 °C.
Mutagenesis, Expression, and Purification of the Cμ4tp Domain.
The Cμ4tp (E446–Y576) gene was optimized for expression in Escherichia coli by GeneArt and cloned into the pET28a expression vector via the NcoI and HindIII restriction sites. Amino acid numbering is according to UniProt entry P01872 (IGHM_MOUSE) with the variable heavy chain (VH) (122 amino acids) added. All mutations on the Cμ4tp domain were introduced using the QuikChange Site-Directed Mutagenesis Kit (Agilent Technologies). Expression was performed, and inclusion bodies were prepared, as well as solubilized, as previously described (13, 42). Insoluble components were removed by centrifugation (46,000 × g, 20 min, 4 °C). The SN was filtered and processed over a HiTrap Q FF column (GE Healthcare) equilibrated with 50 mM Tris (pH 7.5), 10 mM EDTA, and 5 M urea. The protein was collected in the flow-through, diluted to a concentration of 1 mg/mL, and refolded via dialysis into 250 mM Tris⋅HCl (pH 8.0), 100 mM l-arginine, 10 mM EDTA, 1 mM GSSG, and 0.5 mM GSH overnight at 4 °C. Misfolded protein and remaining impurities were removed by processing the sample on a HiLoad Superdex200 26/60 size exclusion chromatography (SEC) column (GE Healthcare) previously equilibrated in PBS buffer. Fractions containing the monomeric Cμ4tp were pooled to obtain the monomer for kinetic experiments. To obtain the hexamer of dimers, the fractions containing the monomer and the dimer were pooled, concentrated, and processed again over the same SEC column. All constructs were sequenced, and the mass of the purified proteins was confirmed using matrix-assisted, laser desorption ionization, time-of-flight mass spectrometry.
Cell Culture, Transfections, Secretion, and Western Blotting.
HEK-293T cells were obtained from the American Type Culture Collection and were cultured in RPMI medium supplemented with 10% FCS, 100 U/mL penicillin, 100 mg/mL streptomycin, and 2 mM l-glutamine (Gibco–Invitrogen). Stable transfectants expressing Ig-λ chains were obtained with pcDNA3.1-λ chain (7) using polyethyleneimine (PEI; Polysciences, Inc.) as previously described (21). Forty-eight hours after transfection, 1 mg/mL G418 was added to the culture medium to select clones with stable insertion of the transgene. As λ is a secreted protein, positive clones were assessed by ELISA of the culture SN. Transient transfection with μ-chain mutants was performed using PEI (21). For analyses of intracellular and secreted IgM mutants, cells were plated in duplicate in six-well plates and transfected. Forty-eight hours after transfection, cells were washed twice with PBS and incubated with minimal essential medium (Opti-MEM) for 4 h. Cell culture SNs were then collected and treated with 10 mM N-ethylmaleimide (NEM; Sigma–Aldrich) to block disulfide interchange (43) and protease inhibitors (Roche). For Western blot assays, cells were detached by trypsinization and washed once in PBS and once in PBS with 10 mM NEM. Cells were then lysed in radioimmunoprecipitation assay (RIPA) buffer [150 mM NaCl, 1% 4-hydroxy-3-nitrophenyl acetyl (Nonidet P-40; Sigma–Aldrich), 0.1% SDS, 50 mM Tris⋅HCl (pH 8.0)] supplemented with 10 mM NEM and protease inhibitors (Roche) for 20 min on ice. To be analyzed by Western blot, SN was subjected to protein precipitation with TCA. IgM in lysates and SN were resolved on SDS/PAGE under reducing or nonreducing conditions. Proteins were then transferred to nitrocellulose and blotted with goat anti-mouse IgM (μ chain) Alexa Fluor 647 (Invitrogen Molecular Probes) and with goat anti-mouse λ-HRP (Southern Biotech). Signals were acquired by Typhoon FLA 9000 (Fujifilm) and Chemidoc Imaging System (UVP).
Sucrose Density Fractionation.
SNs obtained from transiently transfected HEK-293T cells were harvested, and 10 mM NEM was added. After incubation on ice for 20 min, cell SNs were concentrated with TCA, resuspended in 11 mL of homogenization buffer (0.25 M sucrose, PBS, 1% BSA) and fractionated by centrifugation on a continuous sucrose gradient (25–5%) at 273,000 × g in an SW 41 Ti Beckman rotor. Twelve fractions of the gradient were collected. Aliquots from fractions were resolved under nonreducing conditions, transferred to nitrocellulose membrane, and blotted with anti-μ antibody.
Analytical SEC.
SEC-HPLC was performed to determine the oligomeric state of the Cμ4tp wt domain, as well all Cμ4tp mutants in kinetic experiments. For all experiments, a Shimadzu HPLC system was used, and the analysis was performed on a Superdex 200 10/300 GL column (GE Healthcare) in PBS buffer at a flow rate of 0.5 mL⋅min−1 at 20 °C. The absorbance at 280 nm was detected. BioRad gel filtration standard (no. 151-1901) was used as a reference. Experimental details of individual experiments are described in the respective figure legends.
SEC Coupled to Multiangle Light Scattering.
A Tosoh TSKgel G3000SW silica SEC column (7.2 × 300 mm, 10-μm bead, 250-Å pore) and a Shimadzu HPLC system were employed for determination of the absolute mass of oligomers of Cμ4tp mutants. The instrument was coupled to a Wyatt Dawn Helios II multiangle light scattering (MALS) detector, as well as a Shimadzu refractive index and UV detectors. The column was equilibrated for 24 h to obtain stable baseline signals from the detectors before data collection. The interdetector delay volumes and band broadening, the light-scattering detector normalization, and the instrumental calibration coefficient were calibrated using a standard 2-mg/mL BSA solution (Sigma) run in the same buffer, on the same day, according to standard protocols. Protein samples at concentration of 1 mg/mL (20 μL) were loaded on the column. All experiments were performed at room temperature at a flow rate of 0.3 mL⋅min−1 in PBS buffer. The MW and mass distribution of the sample were then determined using the ASTRA 5 software (Wyatt Technology).
CD Spectroscopy.
The secondary structure was determined by CD measurements using a Jasco J-720 spectropolarimeter. Far-UV spectra were recorded from 195 to 260 nm at a protein concentration of 20 μM in 0.5-mm quartz cuvettes. Spectra were accumulated three times and buffer-corrected.
Analytical Ultracentrifugation Sedimentation Velocity Experiments (SV-AUC).
Analytical ultracentrifugation was carried out with a ProteomLab XL-I (Beckman) supplied with absorbance optics. All experiments were performed using PBS at 20 °C. Four hundred fifty microliters of the samples was loaded into assembled cells with sapphire windows and 12-mm path length, charcoal-filled, epon double-sector centerpieces and centrifuged at 42,000–48,000 rpm in an eight-hole Beckman Coulter AN50-Ti rotor. Sedimentation was monitored with an UV/VIS spectrophotometer, equipped with a monochromator, at 280 nm. Data analysis was carried out with the program Sedfit (Peter Schuck, NIH, Bethesda, MD), using a non–model-based continuous Svedberg distribution method, with time and radial invariant noise on.
Resuspension and Oxidation of Peptides.
All peptides were synthesized manually or on an automated synthesizer (Tribute; Gyros Protein Technologies) by Fmoc-based, solid-phase peptide synthesis using 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HBTU) activation in N,N-dimethylformamide (DMF) on Wang resins (44). Double coupling and pseudoproline dipeptide building blocks (Iris Biotech) were used to improve synthesis quality as indicated in Table 2. The final peptide resins were washed with methanol, dried under vacuum, and cleaved with a solution of trifluoroacetic acid (TFA), H2O, and triisoproylsilane (92.5:5:2.5) for 3 h at room temperature. Released peptides were precipitated with cold diethylether (Et2O), washed twice with Et2O, and dissolved in a 1:1 mixture of water and acetonitrile (ACN) containing 0.1% TFA. After freeze-drying, peptides were dissolved in aqueous 6 M Gdn-HCl buffer with 100 mM NaOAc at pH 4. Before final purification, Tris-(2-carboxyethyl)phosphine was added to cysteine-containing peptides to fully reduce them. Purification was achieved by RP-HPLC on a C4 column running a gradient of 5% ACN (ACN + 0.08% TFA) in H2O (+0.1% TFA) to 65% ACN (+0.08% TFA) in H2O (+0.1% TFA). Peptide-containing fractions were identified by electrospray ionization mass spectrometry (ESI-MS) (Waters 3100 Detector) and lyophilized.
Peptide lyophilizates were resuspended in Tris⋅HCl (pH 7.5) and 10 mM EDTA in an appropriate volume to reach the final concentration of 0.4 mg/mL. To oxidize the peptides, 1 mM GSSG and 0.5 mM GSH were added. The samples were then incubated for 2 h at room temperature under stirring.
For material requests or further details on the experimental methods and results, please contact the corresponding authors.
Acknowledgments
This work was supported by grants from the Deutsche Forschungsgemeinschaft (Grant BU836/6-2 to J.B.), Telethon (Grant GGP15059), Ministero della Salute (Grant PE-2011-02352286), Associazione Italiana per la Ricerca sul Cancro (Grant IG 2016-18824), and Fondazione Cariplo (Grant 2015-0591 to R.S.). D.P. was a member of the Graduate School International Max Planck Research School: From Biology to Medicine.
Footnotes
↵1Present address: Interaction Analytics, Protein Sciences & CMC Department, Morphosys AG, 82152 Planegg, Germany.
- ↵2To whom correspondence may be addressed. Email: sitia.roberto{at}hsr.it or johannes.buchner{at}tum.de.
Author contributions: D.P., R.M., C.F.W.B., R.S., and J.B. designed research; D.P., B.W., C.G., T.A., R.M., C.F., M.F., C.J., and M.F.M. performed research; D.P., B.W., and C.G. analyzed data; and D.P., B.W., T.A., R.S., and J.B. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1701797114/-/DCSupplemental.
References
- ↵
- ↵
- ↵.
- Honjo T,
- Alt FW,
- Neuberger MS
- Hendershot LM,
- Sitia R
- ↵
- ↵
- ↵.
- Bornemann KD, et al.
- ↵.
- Fagioli C,
- Mezghrani A,
- Sitia R
- ↵
- ↵.
- Hughey CT,
- Brewer JW,
- Colosia AD,
- Rosse WF,
- Corley RB
- ↵
- ↵
- ↵
- ↵.
- Müller R, et al.
- ↵
- ↵.
- Milstein CP,
- Richardson NE,
- Dieverson EV,
- Feinstein A
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵.
- Sørensen V, et al.
- ↵.
- Anelli T, et al.
- ↵
- ↵
- ↵.
- Randall TD,
- Brewer JW,
- Corley RB
- ↵
- ↵.
- Xia X,
- Longo LM,
- Blaber M
- ↵
- ↵.
- Otzen DE,
- Kristensen O,
- Oliveberg M
- ↵.
- Richardson JS,
- Richardson DC
- ↵
- ↵
- ↵
- ↵.
- Anelli T, et al.
- ↵
- ↵.
- Czajkowsky DM,
- Shao Z
- ↵
- ↵
- ↵
- ↵.
- Anelli T, et al.
- ↵.
- Atherton E,
- Sheppard RC