Throughout the evolution of antibodies from the emergence of IgM in jawed vertebrates 425 million years ago (mya) to the most recent antibody studied here, hominid IgE, which is just 2.5 mya, the homodimeric four-chain structure consisting of two identical heavy (H)-chains and two identical light (L)-chains has been preserved ( Figure 1A ). In this conventionally drawn “Y-shaped” structure, two Fab arms interact with antigen, and the Fc region mediates effector functions such as complement activation and binding to cell surface receptors. Although human IgM consists of either five or, less commonly, six of these H₂L₂ units covalently linked in a (pseudo) hexagonal array ( Figure 1B ), almost all other classes of antibodies function as a single homodimer. The Fc regions of IgM and IgE both consist of six H-chain domains, (C_H2-C_H3-C_H4)₂, as does the evolutionary precursor to mammalian IgE (and IgG), namely avian IgY (1–3), but in other classes of antibodies, such as IgG, IgA, and IgD, an extended “hinge” region replaces the C_H2 domain; this permits greater flexibility between the Fab arms and the Fc region and is important for antigen binding. However, conformational flexibility within the Fc regions of IgM and IgE is known to be critical for expressing their effector functions, but in different ways for these two isotypes, and structures ranging from fully extended to acutely bent have been observed, as described below.

Human IgM, free in serum in the absence of antigen, was shown in early studies to adopt a planar structure, with the Cμ2, Cμ3, and Cμ4 domains in an extended conformation (4, 5) (as shown schematically in Figure 1B ). Both as a pentamer and (even more so) as a hexamer, IgM is a potent activator of complement through the classical pathway, but only in the presence of antigen. Electron micrographs of IgM in the presence of excess antigen show dislocation of the Fab arms from the central Fc disc to form “table” or “staple”-like structures, and these complexes activate complement via the classical pathway by binding of protein C1q (4), the first sub-component of the complement cascade. Feinstein suggested that the C1q-binding sites were obscured in the extended structure and only became accessible in the dislocated bent conformation (4), a proposition supported by later small-angle X-ray scattering (SAXS) analysis and molecular modelling (5). Whether the bend in the IgM-Fc was at the Cμ1-Cμ2 or Cμ2-Cμ3 junction was unclear, but the SAXS analysis suggested the latter (5). Two recent cryo-EM studies of serum IgM (in which the Cμ2 domains are visible), both alone (6) and in the presence of antigen and complement (7), show that there is indeed a bend between the Cμ2 and Cμ3 domains, rather than between the Fabs and the IgM-Fc (i.e., not between Cμ1 and Cμ2hen 2). When IgM is bound to antigen, as in the latter structure, the (Cμ2)₂ domain pair is bent by 100° from its fully extended position with respect to Cμ3 and Cμ4 in the flat disc (7) ( Figure 2A ). Bending of IgM-Fc is an important factor in controlling the complement activity of this antibody. This mechanism is likely preserved through evolution, as jawed fish have multimeric IgM, C1q, and activate complement by the classical pathway (8).

Human IgE-Fc has the same three domain pair composition (Cε2-Cε3-Cε4)₂ ( Figure 1A ), but the crystal structure revealed an acutely bent conformation with the (Cε2)₂ domain pair bent back against the Cε3 domains and even making contact with Cε4 (9). This bent structure also predominates in solution (10–13). The angle between the local two-fold axis of the (Cε2)₂ domain pair and that of the (Cε3-Cε4)₂ pair was 118° in the crystal structure (8) and this increased even further to 126° when the “high-affinity” receptor for IgE, FcεRI, was bound between the two Cε3 domains, accompanied by an opening up of these domains (14) ( Figure 2Bi ). In contrast, when the “low-affinity” IgE receptor FcεRII/CD23 was bound to a site between the Cε3 and Cε4 domains, the bend decreased by 16° and the Cε3 domains adopted a more closed conformation (15) ( Figure 2Bii ). As a consequence of these opposed conformational changes in the positions of the Cε3 domains relative to each other and to the Cε4 domains, the binding of these two receptors is mutually exclusive—an allosteric effect (16); when one site is accessible, the other is not, and vice versa, which is essential for preventing the activation of an allergic reaction via FcεRI by multimeric CD23 in the absence of allergen. IgE-Fc can also adopt a fully extended conformation, which can be trapped and stabilised by anti-IgE-Fc antibody Fabs such as the aεFab (17). However, for free IgE-Fc in solution, this extended conformation, to which FcεRI binding is blocked sterically, may be only rarely populated (17). Bending of IgE-Fc is thus important for the modulation of receptor binding activity.

IgY-Fc also has the same homodimeric composition as IgM-Fc and IgE-Fc, (Cυ2-Cυ3-Cυ4)₂ ( Figure 1A ), but only the crystal structure of the (Cυ3-Cυ4)₂ domains is known (18) and not that of the (Cυ2)₂ domain pair, which may be important in receptor and possible complement binding. This is the first structural analysis of the complete IgY-Fc molecule. One receptor, FcRY, is functionally similar to the mammalian transport receptor FcRn and is involved in the transfer of maternal blood IgY to the yolk and then to the embryo (19); it is known to bind to the Cυ4 domains (20) and also to the Cυ3/Cυ4 interface (19). There is also evidence from the modelling of mutated IgY-Fc interactions with FcRY that changes in the angle between the Cυ3 and Cυ4 domains may occur (19), akin to the conformational changes involving the Cε3 domains of IgE-Fc. Intriguingly, the crystal structure of IgY-Fc exhibits different Cυ3-Cυ4 angles in the two chains (18), further indicating that conformational changes involving these domains may be possible. Another IgY receptor was identified in the chicken genome as homologous to mammalian Fc receptors, FcR/L, but the protein has not been prepared (21). A further two receptors are known, ggFcR (22) (on chromosome 20) and CHIR-AB1 (23) (with a gene in the leukocyte receptor complex), which bind to sites homologous to, firstly, the FcεR1 site on IgE-Fc between the Cυ3 domains (22) ( Figure 2Ci ) and secondly, the FcαR1 site on IgA-Fc, close to the CD23-binding site on IgE-Fc between the Cυ3 and Cυ4 domains (24) ( Figure 2Cii ). In chickens, which have only the IgM, IgA, and IgY isotypes, the latter performs the functions of human IgG (25) and IgE (26, 27), but the functions of the two receptors ggFcR and CHIR-AB1, and whether there is any allosteric communication as in IgE, remain to be investigated. Unlike IgM and IgE, the disposition of the (Cυ2)₂ domain pair of IgY-Fc in solution (or crystal) has not previously been studied.

There is a large gap between the appearance of IgY 310 mya and hominid IgE 2.5 mya, but this is bridged by the monotreme platypus IgE, which appeared 166 mya (28). We have used extant proteins due to their availability and the better, although incomplete, understanding of their function. Our justification for doing so is that the domain structure of all three antibody isotypes has remained constant throughout evolution, and also that for IgM, although its polymeric state has varied, its complement activation function has been conserved.

In this paper, SEC-SAXS (size exclusion chromatography-SAXS) was employed to study the solution conformations of the Fc regions of these four antibodies. In the case of IgM-Fc, the homodimeric subunit of the human pentamer or hexamer was used to permit comparison with the other antibody Fc domains ( Figure 1 ). SEC-SAXS measures the X-ray scattering as the protein is eluted from a size exclusion column, to better resolve monomeric from aggregated material. The scattering data not only permit analysis of the average conformation of each protein in solution but also detect whether more than one, i.e., an ensemble, of different conformations is present. We tracked the evolution of conformational variability and function through this series of antibodies.

Homodimeric Fc regions (C_H2-C_H3-C_H4)₂ of the four antibodies, human IgM, human and platypus IgE, and chicken IgY, were prepared with glycosylation at the conserved “internal” site in C_H3, for analysis of their solution conformation. SEC-SAXS was employed, and the intensity data were analysed using a non-negative linear least-squares algorithm to choose from a pool of models to fit the data and determine whether one or a number of different conformations were present in the solution. The analysis suggests the presence of more than one conformation in each case.

The crystal structure of huIgE-Fc reveals an acutely bent conformation (9), which appears also to predominate in solution, as determined by earlier SAXS (11, 13) and FRET (10, 12) analyses. The energy landscape for IgE-Fc has been explored by molecular dynamics simulations, and an estimate of the energy difference between the acutely bent and fully extended conformations (approximately 20 kJ/mol) suggests that there is little extended IgE-Fc in solution^17. This was confirmed in the present study, with 84% of the material estimated to be in an acutely bent conformation ( Figure 4 ).

In contrast, for huIgM-Fc, the SAXS results show that no acutely bent structures are present, but there are both fully extended and partially bent structures ( Figure 4 ). We have shown, by comparison with the cryo-EM model of the C1q “head” bound to the homologous IgG-Fc (PDB 6FCZ) as described above ( Supplementary Figure 3 ), that in 36% of these structures, the C1q binding site is accessible. However, free pentameric IgM does not bind C1q or activate complement, suggesting that the formation of the pentamer (or hexamer) may affect the conformational diversity in IgM-Fc. Although pentameric IgM (6) and IgM-Fc (43–45) cryo-EM structures imply that flexibility and bending at the Cμ2-Cμ3 interface can occur, even in the absence of antigens, the extent of this bending is not known. Other factors, such as glycosylation, may also affect the ability of IgM-Fc to bend and activate complement. A recent study showed that in the sera of patients with a severe case of COVID-19, increased glycosylation of IgM-Fc correlated with increased IgM-dependent complement deposition (46). We note that in the recently determined structures of the IgM B-cell receptor (33, 34, 47), the Fc region adopts an extended conformation, perhaps stabilised by the accessory Igα and Igβ molecules.

IgE-Fc from platypus, the most evolutionarily primitive of the extant mammals, was included in this study to offer an evolutionary intermediate between chicken IgY-Fc and human IgE-Fc. In fact, the profile of conformations seen for plIgE-Fc is very similar to that of huIgE-Fc, dominated by an acutely bent conformation (67%, Figure 4 ). CD23 has been annotated in the platypus genome (28), and fragments of an FcεR1α homologue have been identified with Blast (48). However, key residues in the huIgE-Fc-binding site for CD23 (15, 16) are not conserved in plIgE-Fc, and binding of platypus FcεR1 to plIgE-Fc has never been demonstrated, nor have there been reports of an allergic reaction, although one individual of the only other extant monotreme, the echidna, was found to be allergic to its main source of food, namely ants (49). The presence of IgE, CD23, and possibly FcεR1α in the platypus genome suggests that the system seen in humans (50) had already evolved to combat helminths that infect monotremes and marsupials (51).

Acutely bent and fully extended structures occur in the ensemble for chIgY-Fc, together with intermediate conformations, which are clearly different from the distributions seen for either IgM-Fc or IgE-Fc ( Figure 4 ). The classical complement pathway has been found in chicken (52), but its activation could be due to IgY or IgM. There are no structural or mutational data to indicate how chicken IgY might interact with chicken complement, but two out of the three important residues in both the human IgG and IgM interaction with human complement, P329 and P331, align with prolines in IgY. Chicken IgY only exists as a homodimer (in contrast to the pentameric or hexameric IgM), and the SAXS results show that 31% of it [as determined by performing the same superposition of the IgG-Fc/C1q complex (42) (PDB 6FCZ) as described above for IgM-Fc] is bent sufficiently to expose the complement binding site if it exists; the disposition of the (Cυ2)₂ pair is, therefore, unlikely to affect activation. The range of conformations for chicken IgY-Fc and the putative C1q-binding site is illustrated in Supplementary Figure 4 . If IgY can activate chicken complement, the formation of polymeric structures upon antigen binding is likely to be the control mechanism, as it is with IgG (53, 54), because enough of the IgY-Fc is sufficiently bent to allow C1q to bind. Although chickens do experience anaphylaxis (26), there is no evidence that IgY and a receptor are involved. On the other hand, it is known that IgY has some functions analogous to those of both human IgG and IgE, which may involve the identified receptors ggFcR (22) and CHIR-AB1 (23), which bind to IgY-Fc at similar locations, respectively, to FcεRI and CD23 on IgE-Fc (22, 24) ( Figures 2B, C ). The presence of more than one bent conformation suggests that, like huIgE-Fc, different dispositions of the (Cυ2)₂ domain pair may correspond to the binding of different chicken receptors, and changes in the Cυ3-Cυ4 angle may also occur (18, 19), as described in the Introduction, similar to those that are associated with the open and closed conformations of the Cε3 domains in IgE-Fc (14, 15). Allostery in relation to IgY receptor binding might therefore be possible, as it is for IgE.

We have studied and compared the solution structures of human IgM-Fc, platypus and human IgE-Fc, and chicken IgY-Fc, the latter being the first structural study of the complete Fc region of this isotype. While IgM-Fc is principally fully extended in solution, with no acutely bent conformations, IgE-Fc is principally acutely bent in solution, with no fully extended conformations. In contrast to both IgM-Fc and IgE-Fc, IgY-Fc displays an ensemble of conformations ranging from acutely bent to fully extended, reflecting IgY’s position as an evolutionary intermediate between IgM and IgE. Indeed, the proportion of acutely bent conformation, IgM-Fc < IgY-Fc < plIgE-Fc < huIgE-Fc, follows their order of appearance in evolution. The solution structure of human IgM-Fc (which was already present 425 mya in jawed fish) is consistent with IgM using the bend between the (Cμ2)₂ domain pair and (Cμ3-Cμ4)₂, influenced by glycosylation and/or multimerisation, to control complement activation upon antigen binding. Hominid IgE (which only appeared 2.5 mya) does not bind complement, but the bend between (Cε2)₂ and (Cε3-Cε4)₂ is associated with the mutually exclusive allosteric binding of two receptors, FcεRI and CD23. This function probably also exists in the monotremes platypus and echidna (which first appeared 166 mya). The solution structure of IgY-Fc (which appeared 310 mya in tetrapods) suggests that it has lost the ability to control complement binding using the (Cυ2)₂ domain pair and may have already co-opted that binding site for Fc receptor binding, opening up the possibility of allosteric control of receptor binding as seen in human IgE.

In summary, IgY occupies an intermediate position in the evolution of antibody structure from the most primitive IgM to the most recent IgE. This is the first structural analysis of the complete IgY-Fc that allows direct comparison with IgM-Fc and IgE-Fc; we found that IgY-Fc displays a unique ensemble of conformations in solution, distinct from IgM-Fc and IgE-Fc but displaying features of both. We anticipate that this unique structure of IgY-Fc will be reflected in unique biology and function, in particular with respect to its receptor interactions, further illustrating the co-evolution of antibody structure and function.