Designed IgM from glycoengineering

Among the immunoglobulins, the IgGs have received the lion’s share of attention and dominate the monoclonal antibodies (mAbs) that are US Food and Drug Administration-approved or are in research pipelines leading to the clinic. Whereas IgG is relatively easy to produce and purify and a number of host expression systems have been developed for that isotype, interest in the utility of other isotypes has been increasing in the last few years. In particular, IgM has begun to receive considerable attention as an isotype that may be ideal for specific clinical indications (1). IgM differs from IgG in a number of respects. First and foremost, IgMs are extraordinarily complex molecules consisting of an array of 10 (pentameric IgM) or 12 (hexameric IgM) combining sites compared with the 2 combining sites that comprise IgG. This extensive array consists of heavy chains (μ), light chains (predominantly κ), and may have J chains covalently attached to the pentameric form. As a consequence, IgMs are enormous molecules (∼900 kDa as opposed to ∼150 kDa for IgG). This complexity comes with certain clinically relevant features. In particular, the high level of valency of IgMs can give rise to a very significant degree of avidity for antigen. This is in contrast to IgG, where affinity for antigen is key, and avidity plays a secondary role. Because of the high avidity and relatively low affinity, IgM can be polyreactive. Because a significant proportion of IgM is germ line-encoded and emerges during embryogenesis without apparent antigenic stimuli, they have been referred to as “natural antibodies” (2). Indeed they are the first isotype to be expressed during embryogenesis, as well as the first isotype of an immune response to antigen (3). Due to the abundance of Fc regions, IgM can very efficiently activate complement, leading to potent cytotoxic and cytolytic activity. A single IgM molecule is capable of activating complement. Activation by IgM can therefore be about 1,000 times more efficient than IgG (3, 4). IgM in human serum (∼5% of total Ig) consists predominantly of the pentameric form with only a trace of the hexameric form. Different species have different pentamer:hexamer ratios, but invariably the two forms of IgM are functionally distinct, with the hexameric form being significantly more efficient in complement activation (5). We don’t yet understand the factors that control the pentamer:hexamer ratio or why only ∼50% of pentameric IgM contains a J chain. An additional component of IgM complexity is glycosylation. IgG ordinarily has a single glycosylation site on each constant region, whereas pentameric IgM has 10 sites at five different locations in the constant region. Also, whereas we are beginning to understand how glycosylation can affect

Given the results reported by Loos et al., we now have the opportunity to investigate the functional consequences of glycan manipulation of IgM.

the biological properties of IgG, we have no such understanding of IgM glycosylation. For this reason, there has been a keen interest in developing host systems that can be used for glycoengineering and efficient expression of monoclonal IgM variants. One such system is described in PNAS by Loos et al. (6).

The Plant Production System

In the paper by Loos et al., the assembly of both pentameric and hexameric forms of IgM (in approximately equal proportion) is demonstrated using a Nicotiana benthamian plant system. This transient system has been gaining popularity for antibody expression due to the rapid turn-around time from recombinant gene to purified antibody (7, 8). Requiring less than 2 wk for high level expression to occur, the system can readily be used for the iterative process of genetic manipulation and functional analysis. The N. benthamiana used for IgM expression in Loos et al. is itself a transgenic plant that uses RNA interference technology (RNAi) for the targeted reduction of the endogenous plant xylosyl- and fucosyl-tranferase enzymes (referred to as a ∆XF plant). These enzymes attach their respective sugars to the glycan core via nonmammalian β1,2-xylosyl and α1,3-fucosyl linkages. Previous work by this group, using IgG mAb, demonstrated that the antibody glycans resulting from expression in the ∆XF host can be remarkably homogeneous, consisting primarily of GlcNac₂Man₃GlcNac₂ (GnGn). Because the plant-specific linkages are eliminated, this glycan is indistinguishable from a mammalian glycan. By virtue of lacking fucose, this glycan confers significantly enhanced FcγRIII binding and enhanced antibody-dependent cellular cytotoxicity (ADCC) activity to IgG (9). Moreover, this core glycan can further serve as the substrate for additional modifications including mammalian galactosylation and sialylation, both of which can have significant functional consequences on IgG antibodies (10, 11). We have no such appreciation for the impact of glycan structure on IgM function. The paper by Loos et al. provides a means to begin that evaluation by creating a variety of IgM glycoforms available for functional analysis.

IgM Glycosylation

The authors first demonstrate that glycosylation of human serum IgM is site specific. That is, each of the five glycosylation sites has a different glycan profile. This is especially true when comparing sites 1–3 with sites 4 and 5. Sites 1–3 are dominated by fucosylated, monosialylated complex glycans, although some disialylated glycans are observed. Sites 4 and 5 are comprised entirely of oligomannosidic structures. These results are consistent with previously published analyses of human serum IgM glycosylation (12). The authors then expressed the IgM in both ∆XF plants, as well as ∆XF plants containing the enzymes for sialylation (13). Because plants do not contain sialic acid, an extensive mammalian enzyme coexpression strategy was required to achieve sialylation. This involved introduction of uridine diphosphate-N-acetylglucosamine 2-epimerase, N-acetylneuraminic acid phosphate synthase, β1,4-galactosyltransferase, and α2,6-sialyltransferase. Such extensive coexpression is facilitated by the high efficiency of cellular coinfection of N. benthamiana using multiple recombinant Agrobacteria, each delivering the genetic elements for expression of a particular enzyme (13). The resulting IgM with mammalian-like glycans demonstrates that glycoengineering of IgM can be accomplished in a site-specific manner using the N. benthamiana system. In particular, greater than 50% of the glycans were of the GnGn variety when IgM was simply expressed in the ∆XF plant. When sialylation enzymes were present, significant sialylation (40%) was observed in sites 1–3, of which 17% were found to be disialylated. No sialylation occurred in sites 4 or 5. As with other host expression systems, the 50:50 ratio of pentameric to hexameric IgM species was observed despite the presence of the J chain. In all cases, the antigen-binding capability of the IgM was unaltered, and the overall molecular structure was not significantly different from mammalian cell line-derived IgM. Moreover, based on previously generated structural data, the authors generated a computer model of IgM and its glycosylation. The resulting model suggests that one explanation for the differential glycosylation of sites 1–3 vs. sites 4 and 5 may have to do with the inaccessibility of sites 4 and 5, which are located in the region where Fc backbones are highly clustered.

IgM Receptors

Our understanding of the various receptors that are capable of binding IgM is relatively sparse compared with the wealth of information we have about the FcγR receptor varieties. We know that some IgM species bind to the polyimmunoglobulin receptor by virtue of having a covalently attached J chain. As such, IgM has a significant distribution on mucosal surfaces, a characteristic shared with J chain-containing dimeric IgA. Another shared feature with IgA is the binding to the Fcα/μ receptor, which appears to be involved in the primary stages of the immune response to bacterial pathogens by mediating primary B-lymphocyte endocytosis of IgM-coated microbes (14). A specifc IgM receptor has only been identified within the last 5 y (15). It appears to play a critical role in IgM homeostasis, B-cell survival, and humoral immune responses (1), although this is an emerging field of investigation. Given the results reported by Loos et al., we now have the opportunity to investigate the functional consequences of glycan manipulation of IgM. This has the potential to jump start research into the structures of IgM that are most appropriate for high level production and ultimately for clinical development. A number of anticancer IgMs have been isolated in the last two decades, some of which are specific for malignant tissues and mediate tumor-specific apoptosis (16, 17). It is a timely development that we now have the tools to explore glycoengineered IgM variants that may have enhanced clinical efficacy.